mrkeyoor.com_
Sat 31 Oct 21:52 UTC

Best Open-Source Workflow Automation Tools

Self-hostable alternatives to Zapier and friends, ranked by how honest the setup story really is. Ranked by our usability scores across setup ease, docs, community health, and maturity — 255 tools reviewed in full.

01

homebrew-cask

Homebrew/homebrew-cask · ★ 22k ▲5/7d · Ruby

Homebrew Cask is the catalogue that lets Homebrew install and manage prebuilt applications, fonts, plugins, and other vendor packages. It replaces manual download-and-drag installation with repeatable `brew` commands, while each cask records where an artifact comes from and how Homebrew should install or remove it.

SetupDocsCommunityMaturity

Verdict GitHub listed 29 open issues and pull requests and a push on August 25, 2026, so Homebrew Cask is an actively maintained catalogue despite its 2016 release tag. It is the default choice for scriptable Mac application setup when Homebrew is already present, but it is not a fleet-management system and it cannot make inconsistent vendor installers uniform.

Full review →
02

checkout

actions/checkout · ★ 9.0k ▲15/7d · TypeScript

Use Actions Checkout v7 for GitHub Actions jobs that need repository files, provided your runner meets its requirements and you review history and credential settings. Our run on October 2, 2026 tested commit f548e57 in our sandbox; it did not exercise a deployed GitHub Actions workflow. Actions Checkout places repository files under GITHUB_WORKSPACE for later workflow steps.

SetupDocsCommunityMaturity

Verdict Actions Checkout v7 is a reasonable choice for GitHub Actions jobs that need repository files, after checking runner requirements, history depth and credentials. Our October 2, 2026 sandbox run at f548e57 completed the repository commands successfully, but it did not test a deployed workflow.

Full review →
03

actionlint

rhysd/actionlint · ★ 4.4k ▲19/7d · Go

Use actionlint to check GitHub Actions workflows before running them, but keep workflow execution and security review as separate checks. actionlint is a static checker for GitHub Actions workflow files. Our run on October 1, 2026, at commit 011a6d1 recorded a successful build step and a failed test command; it did not establish that a deployment or representative workflow worked.

SetupDocsCommunityMaturity

Verdict Add actionlint as an early check for GitHub Actions workflow definitions if that is the gap you need to fill. Our October 1, 2026, sandbox run at commit 011a6d1 installed and built successfully, but its test command failed.

Full review →
04

yt-dlp

yt-dlp/yt-dlp · ★ 204k ▲1.9k/7d · Python

yt-dlp downloads video and audio from websites through a command-line tool or Python API. It helps you repeat downloads, select media formats, and keep subtitles or metadata with the saved files.

SetupDocsCommunityMaturity

Verdict Our yt-dlp run installed in 13 seconds and built in 2 seconds, but the suite ended with 200 failures alongside 318 passes, so it earns a trial on your target sites rather than blind trust in scheduled jobs. Use it when repeatable downloads and detailed format control matter enough to maintain the external tools and update policy.

Full review →
05

ansible

ansible/ansible · ★ 71k ▲48/7d · Python

Ansible runs repeatable infrastructure changes from one control machine, usually over SSH, so managed Linux hosts do not need a resident Ansible agent. This repository contains the ansible-core engine and built-in content; the larger ansible package adds separately packaged modules called collections for more systems and services.

SetupDocsCommunityMaturity

Verdict Our ansible/ansible run installed 42 packages in 28 seconds and built in 6 seconds, then pytest stopped with 7 collection/setup errors before running a test. The released tool is easy to try, while contributor work needs Ansible's documented test tooling and a closer look at its Python setup.

Full review →
06

Win11Debloat

Raphire/Win11Debloat · ★ 61k ▲478/7d · PowerShell

Win11Debloat is a PowerShell script that removes preinstalled Windows 11 apps, disables telemetry and ads, and applies dozens of UI and privacy tweaks in one run. It solves the problem of manually clicking through Windows settings or hunting down app removal commands.

SetupDocsCommunityMaturity

Verdict We could not run it: no PowerShell in our Debian sandbox and no Dockerfile, so there are no install/build/test numbers from us. For a Windows user, Win11Debloat is a proven, active, and very popular way to strip Windows 11 down to a cleaner state in minutes, but you are trusting a script with administrator rights to do a lot at once..

Full review →
07

helm

helm/helm · ★ 30k ▲13/7d · Go

Helm is a package manager for Kubernetes applications. It bundles configurable Kubernetes manifests into charts, installs them through the cluster API, and keeps release history so operators can upgrade, inspect, roll back, or remove an application as one named unit.

SetupDocsCommunityMaturity

Verdict Our Helm run installed 341 packages in 46 seconds, built successfully in 178 seconds, and passed all 118 tests in 151 seconds. Use Helm 4 when reusable application packaging and release history justify a template layer in front of Kubernetes.

Full review →
08

celery

celery/celery · ★ 29k ▲29/7d · Python

Celery is a Python task queue that sends work from an application through a message broker to worker processes. It moves slow, scheduled, or retryable jobs out of a web request and can route them across many workers while optionally recording their status and results.

SetupDocsCommunityMaturity

Verdict Our Celery run passed 3,912 tests and failed 8 privilege-check cases in 599 seconds, which shows a deep test suite and a real environment-sensitive edge. Use Celery when background work has grown into shared infrastructure and your team can own a broker, worker supervision, delivery semantics, and monitoring.

Full review →
09

jenkins

jenkinsci/jenkins · ★ 27k ▲18/7d · Java

Jenkins is a self-hosted automation server used to build, test, analyze, and deploy software. Its controller schedules work across local or remote executors, while more than 2,000 plugins connect jobs to source control, build tools, credentials, notifications, and deployment systems.

SetupDocsCommunityMaturity

Verdict Our Jenkins frontend run installed 715 packages in 57 seconds, then built successfully and passed all 5 Vitest tests. Jenkins remains a sensible choice for mixed infrastructure and established plugin-heavy pipelines when a team can own the controller, agents, credentials, and upgrade testing.

Full review →
10

pulumi

pulumi/pulumi · ★ 26k ▲22/7d · Go

Pulumi is an infrastructure-as-code system that lets teams define cloud resources with TypeScript, JavaScript, Python, Go, .NET, Java, or YAML. Its CLI evaluates that program, compares the desired resources with stored stack state, and asks cloud providers to create, update, or delete infrastructure.

SetupDocsCommunityMaturity

Verdict Our `npm/` run passed 15 of 15 tests in 14 seconds and found 0 known vulnerabilities, but it covered only a tiny corner of Pulumi's 1.7-million-line repository. Pulumi is a strong choice when developers will genuinely benefit from normal language features and the organization can govern state, credentials, plugins, and previews.

Full review →
11

argo-cd

argoproj/argo-cd · ★ 25k ▲30/7d · Go

Choose Argo CD when your Kubernetes team wants Git-defined deployment state with an application interface and can take responsibility for cluster credentials and permissions. Argo CD is a declarative GitOps continuous delivery tool for Kubernetes. Our run on October 1, 2026, at commit 519e79f recorded a successful build step and a failed test step. That sandbox result does not establish that an application was deployed or reconciled.

SetupDocsCommunityMaturity

Verdict Evaluate Argo CD if Kubernetes deployment state belongs in Git and your team wants its application interface and access controls. Our October 1, 2026, run at commit 519e79f did not produce a clean test result, and it did not demonstrate a working deployment.

Full review →
12

renovate

renovatebot/renovate · ★ 23k ▲82/7d · TypeScript

Renovate finds dependency references in a repository, checks for newer releases, and opens pull requests with the version and lockfile changes. It covers more than 90 package managers and several source-control platforms, giving teams one policy engine for updates that would otherwise be checked by hand.

SetupDocsCommunityMaturity

Verdict Our Renovate run installed 1,094 packages and built in 15 seconds, but its test suite ended in failure after 607 seconds, so platform teams get deep dependency automation with a substantial codebase behind it. Choose Renovate when more than 90 package managers, private registries, grouping, and scheduled update policy need to work the same way across many repositories.

Full review →
13

claude-code-action

anthropics/claude-code-action · ★ 9.7k ▲81/7d · TypeScript

Claude Code Action puts Claude inside GitHub issues and pull requests, where it can answer questions, review changes, edit code, and run prompted maintenance jobs. It solves the handoff between a repository conversation and an agent working on the checked-out code.

SetupDocsCommunityMaturity

Verdict Our run installed 150 packages in 31 seconds and all 966 tests passed, which makes the action easy to trust at the repository-mechanics level. Use it when Claude is already an approved provider and maintainers will keep triggers, tools, and GitHub permissions narrow.

Full review →
14

App-Store-Connect-CLI

rorkai/App-Store-Connect-CLI · ★ 8.1k ▲106/7d · Go

Consider App Store Connect CLI (`asc`) if you want scriptable Apple release workflows, but verify your release path before adopting it: our run on October 2, 2026, at commit 8ba978a ended with a nonzero test exit. App Store Connect CLI automates App Store Connect workflows from a terminal or CI pipeline. It is an independent, unofficial tool; our sandbox did not verify an authenticated Apple release.

SetupDocsCommunityMaturity

Verdict App Store Connect CLI is worth evaluating for command-driven Apple release work, with a workflow-specific acceptance check before adoption. Our run on October 2, 2026, at commit 8ba978a had successful install and build steps but a failing test step.

Full review →
15

puppeteer

puppeteer/puppeteer · ★ 96k ▲29/7d · TypeScript

Puppeteer is a TypeScript and JavaScript library for controlling Chrome or Firefox from Node.js. It turns navigation, form input, screenshots, PDF creation, and page inspection into code through the DevTools Protocol or WebDriver BiDi.

SetupDocsCommunityMaturity

Verdict Our Puppeteer run built in 68 seconds, then finished with 9 passing and 144 failing tests, so adopting it means validating your exact browser host instead of trusting a successful compile. Use it for Node.js jobs that need close control of Chrome and can pin the browser, cache, Linux packages, and sandbox setup.

Full review →
16

ai-job-search

MadsLorentzen/ai-job-search · ★ 68k ▲2.3k/7d · Python

AI Job Search turns a local repository into a Claude Code workspace for finding openings, judging fit, tailoring a CV and cover letter, and preparing for interviews. It keeps the working files under your control and includes portal tools for Denmark plus broader LinkedIn and FreeHire searches.

SetupDocsCommunityMaturity

Verdict Our run installed 35 packages in 15 seconds and pytest reported 494 passed with 0 failures, so AI Job Search has unusually credible engineering underneath its prompt files. Use it if you already pay for Claude Code and want a careful, local application process you can inspect.

Full review →
17

rclone

rclone/rclone · ★ 61k ▲115/7d · Go

Rclone is a command-line tool for copying and synchronizing files between cloud storage accounts, servers, and local folders. It gives you one set of commands for moving files across providers instead of managing a separate client for each service.

SetupDocsCommunityMaturity

Verdict Our rclone run built successfully in 176 seconds but finished its test step with 223 passing results and 15 failures, so this checkout needs closer inspection before a patched release. Use it for deliberate transfers between storage providers when you can configure credentials and verify the destination.

Full review →
18

huginn

huginn/huginn · ★ 50k ▲84/7d · Ruby

Choose Huginn for self-hosted monitoring if you can maintain its database, web process, and background worker. Its agents pass events through a directed graph to connect monitoring with actions. Our run on 2026-10-02 covered commit 86ac281 through the Node harness; it did not verify a running Rails application or an agent workflow.

SetupDocsCommunityMaturity

Verdict Evaluate Huginn when self-hosted monitoring justifies maintaining a database, web process, and worker. Our Node run completed installation and the URL-polyfill build, skipped tests, and recorded one high-severity npm advisory.

Full review →
19

airflow

apache/airflow · ★ 47k ▲52/7d · Python

Apache Airflow schedules and monitors workflows whose steps and dependencies are written in Python. It gives recurring, mostly stable jobs a shared scheduler, run history, retries, and a web interface, replacing scattered cron entries with workflow code a team can review.

SetupDocsCommunityMaturity

Verdict Our Airflow run installed 152 packages in 94 seconds and built in 21 seconds, but pytest stopped after 7 seconds on an unrecognized option, so the source tree expects more preparation than the package quick start suggests. Use Airflow when scheduled Python DAGs, backfills, provider integrations, and years of run history justify shared infrastructure.

Full review →
20

agent-browser

vercel-labs/agent-browser · ★ 45k ▲350/7d · Rust

agent-browser is a command-line browser controller built for coding agents. It turns a live page into short accessibility references, then lets an agent click, fill forms, inspect state, take screenshots, or connect through MCP without writing a browser driver first.

SetupDocsCommunityMaturity

Verdict Our install of commit 39a74c7 took 138 seconds, pulled 1,097 packages, and left 931 MB on disk, while the checkout exposed no build or test target. Use agent-browser when a coding agent needs concise Chrome actions, fresh page references, and a typed MCP path.

Full review →
21

fastlane

fastlane/fastlane · ★ 42k ▲23/7d · Ruby

fastlane is a Ruby command-line tool that automates tedious iOS and Android release tasks like generating screenshots, managing provisioning profiles, and uploading apps. It solves the real problem of Apple code signing and manual App Store or Play Store submission eating days of developer time.

SetupDocsCommunityMaturity

Verdict fastlane has 42,092 GitHub stars, a last push one day ago, and a 2.239.0 release six days before this review, but our run measured zero install, build, or test steps because the repo has no supported ecosystem. That makes it hard to verify setup ease from our sandbox, though the project's longevity and active maintenance suggest it remains the default mobile release automation tool for teams that already run Ruby..

Full review →
22

prefect

PrefectHQ/prefect · ★ 24k ▲28/7d · Python

Prefect is a Python workflow system that adds scheduling, retries, caching, state tracking, and a web dashboard to ordinary functions. It helps data teams run scripts repeatedly, see what failed, and move the same workflow from a laptop to workers without rewriting it as a separate configuration language.

SetupDocsCommunityMaturity

Verdict Our Prefect run installed 128 packages in 53 seconds and built in 12 seconds, but tests stopped before collection because `numpy` was missing. Prefect is a strong choice when Python workflows have outgrown cron and a team will use the run history, retries, parameters, and worker model.

Full review →
23

temporal

temporalio/temporal · ★ 24k ▲116/7d · Go

Temporal is a durable execution server for application workflows that must survive process crashes, network failures, retries, and long waits. Developers write workflow and activity code with a language SDK, run their own workers, and let the server persist event history and coordinate the next task.

SetupDocsCommunityMaturity

Verdict Our Temporal build took 278 seconds, and 13 of 15 test targets passed before two PostgreSQL CLI tests failed, so source contributors should expect a large server with persistence-sensitive tests. Temporal is worth the conceptual and operating cost when a failed business process cannot simply be restarted from step one.

Full review →
24

Karabiner-Elements

pqrs-org/Karabiner-Elements · ★ 23k ▲29/7d · C++

Karabiner-Elements is a macOS keyboard remapper that can make one key act like another or apply different rules by device, application, and input state. It solves awkward keyboard layouts and repetitive shortcuts at a lower level than most app-specific hotkey settings.

SetupDocsCommunityMaturity

Verdict Our sandbox found 25 known npm vulnerabilities in the vendored Duktape debugger and could not build or test the macOS app, so the 42-second install is not evidence that Karabiner-Elements itself passed a health check. The official binary remains a strong choice for Mac users who need device-aware remapping and can accept its deep system permissions.

Full review →
25

appium

appium/appium · ★ 22k ▲41/7d · TypeScript

Appium is a test-automation server that lets one style of test control native, hybrid, mobile-web, and desktop apps through installable platform drivers. It gives teams a common interface across device families while letting each driver handle the platform-specific work.

SetupDocsCommunityMaturity

Verdict Our Appium run passed all 416 tests after a 61-second install, but it left 573 MB of dependencies and 14 high-severity audit findings, so it earns its keep only when one automation protocol must cover several app platforms. Use it for mixed Android and iOS estates, multi-language test teams, or device-cloud infrastructure.

Full review →
26

browser-harness

browser-use/browser-harness · ★ 19k ▲157/7d · Python

Browser Harness lets a coding agent control a real Chrome session through the Chrome DevTools Protocol. It can use existing logins, click and type on interactive pages, and write reusable Python helpers when a site needs behavior the base tool does not provide.

SetupDocsCommunityMaturity

Verdict Our Browser Harness install, build, and all 140 tests passed in 61 seconds combined, with 0 known package vulnerabilities. It is a strong fit for a careful developer who wants an agent to use a real logged-in Chrome session and can watch high-impact actions.

Full review →
27

windmill

windmill-labs/windmill · ★ 18k ▲64/7d · Rust

Choose Windmill when developers own the automation and colleagues need interfaces to run it. Windmill turns scripts into generated interfaces and lets teams compose them into workflows and internal apps. Our run covered CLI installation, build and tests, so it does not establish that a self-hosted server or a production workflow works.

SetupDocsCommunityMaturity

Verdict Windmill is worth evaluating when developer-owned scripts also need workflows and internal interfaces. Our October 2, 2026 CLI run at `fee401f` installed and built successfully, but tests stopped with a Cargo build error during setup.

Full review →
28

keda

kedacore/keda · ★ 11k ▲19/7d · Go

Choose KEDA when Kubernetes workloads need event-driven scaling, including scaling to and from zero, and your team can validate that behavior against its event source. KEDA integrates with the Horizontal Pod Autoscaler rather than replacing it. Our run on October 1, 2026, at commit 81c6b83 recorded successful install and build steps, but the test command failed. It did not establish that a deployed workload could scale correctly.

SetupDocsCommunityMaturity

Verdict KEDA is worth evaluating when event-driven Kubernetes scaling, including activation from zero, is the requirement. Our October 1, 2026 run at commit 81c6b83 completed the recorded install and build steps but failed its test command.

Full review →
29

mobile-mcp

mobile-next/mobile-mcp · ★ 10k ▲176/7d · TypeScript

Mobile MCP is a Model Context Protocol server that lets an AI agent inspect and control iOS and Android apps through one set of tools. It uses device accessibility data where possible, with screenshots and coordinate taps as a fallback, so agents can explore apps without separate XCUITest and Espresso scripts.

SetupDocsCommunityMaturity

Verdict Our Mobile MCP run installed 437 packages in 17 seconds, passed its build and tests in 19 seconds combined, and reported 0 known npm vulnerabilities, making it an easy MCP server to trial. Use it for agent-led inspection and exploratory mobile flows when you already have the platform SDK and device setup.

Full review →
30

lego

go-acme/lego · ★ 10.0k ▲19/7d · Go

Lego is a Go command-line client and library for obtaining, renewing, and revoking TLS certificates from Let's Encrypt and other ACME certificate authorities. Its main draw is DNS challenge automation for more than 200 providers, alongside HTTP and TLS challenge support.

SetupDocsCommunityMaturity

Verdict Our lego install never completed: after 682 seconds, Go module downloads ended on `proxy.golang.org` stream errors before build or tests began. Use a v5.4.1 release binary or pinned container when your DNS provider is supported, then test that provider against a staging CA before touching production records.

Full review →
31

atlantis

runatlantis/atlantis · ★ 9.3k ▲5/7d · Go

Atlantis is a self-hosted service that runs Terraform plans and applies from pull-request comments. It puts infrastructure changes, plan output, approvals, and execution history in the same review thread instead of relying on engineers to run Terraform from their laptops.

SetupDocsCommunityMaturity

Verdict Our Atlantis npm run installed 300 packages and found 0 known vulnerabilities, but it exposed no build or test target, so the 157,048-line checkout did not receive an executable server-level check from this harness. Use Atlantis when pull requests are already the control point for Terraform and you can treat the service as privileged infrastructure.

Full review →
32

gogcli

openclaw/gogcli · ★ 8.5k ▲16/7d · Go

gogcli is one command-line client for Gmail, Calendar, Drive, Docs, Sheets, and much of Google Workspace. It gives people, scripts, and agents stable machine output, explicit account selection, and controls that can prevent an automated caller from sending mail or changing cloud data.

SetupDocsCommunityMaturity

Verdict Our gogcli run installed 139 packages, built successfully, and passed all 84 tests, so the code is a credible base for Google Workspace automation. Choose it when one explicit, scriptable interface across several Google services is worth the OAuth setup, especially for read-only jobs or tightly allowed agent tools.

Full review →
33

hatchet

hatchet-dev/hatchet · ★ 8.2k ▲24/7d · Go

Consider Hatchet when application jobs need retained execution history, retries and durable waits, and your team can evaluate the service behind those features. Hatchet supports Python, TypeScript, Go and Ruby applications through Hatchet Cloud or self-hosting. Our run on October 1, 2026, at commit 495421a recorded a failed test command; it did not establish production readiness or exercise a deployed application workflow.

SetupDocsCommunityMaturity

Verdict Pilot Hatchet if retained workflow history and recovery controls address a concrete application need. Our October 1, 2026 run at commit 495421a leaves adoption work: resolve the database-container requirement, investigate the remaining test failures and check the vet warning.

Full review →
34

karpenter-provider-aws

aws/karpenter-provider-aws · ★ 7.7k ▲7/7d · Go

Karpenter Provider AWS is a Kubernetes controller that launches suitable EC2 instances when pods cannot be scheduled, then removes or replaces nodes as demand changes. It chooses capacity from pod requirements and AWS instance options instead of making teams predefine a node group for every workload shape.

SetupDocsCommunityMaturity

Verdict Our Karpenter build passed in 180 seconds, but tests ended with 19 of 21 packages failing, so source contributors need the project's intended test environment before treating a generic Go run as healthy. For EKS teams, Karpenter is a strong choice when mixed workloads make fixed node groups wasteful and operators can govern IAM plus disruption.

Full review →
35

gh-aw

github/gh-aw · ★ 5.5k ▲65/7d · Go

Consider GitHub Agentic Workflows for repository tasks that need interpretation, provided someone owns permissions and reviews the results. GitHub Agentic Workflows compiles Markdown instructions into standard GitHub Actions workflows. Our run on October 2, 2026, at commit 35eb607 recorded successful install and build steps but a failed test command; it did not test a deployed agent workflow.

SetupDocsCommunityMaturity

Verdict Evaluate gh-aw for supervised repository tasks that need interpretation. Our sandbox established successful install and build command outcomes, followed by test failure; it did not establish a working deployed agent.

Full review →
36

n8n

n8n-io/n8n · ★ 209k ▲563/7d · TypeScript

n8n is a visual builder for automated workflows and AI agents, with code steps for jobs that do not fit a prebuilt node. You can use its hosted service or run it yourself when workflows, credentials, and execution data need to stay under your control.

SetupDocsCommunityMaturity

Verdict Our n8n run installed 3,408 packages and built successfully, but the full test command exited 1 after the CLI package reported 1 failed test. Use it when 1,500-plus integrations and a visual canvas can retire enough glue code to justify operating a 3,296 MB source environment.

Full review →
37

remotion

remotion-dev/remotion · ★ 69k ▲1.7k/7d · TypeScript

Remotion turns React components into videos and images, so layouts, timing, data, and brand rules can live in code. It gives developers a Studio for authoring plus local, server, cloud, and browser rendering paths for repeated video work.

SetupDocsCommunityMaturity

Verdict Our Remotion install failed after 206 seconds when Electron's postinstall timed out connecting to 185.199.109.133:443, so repository work needs a dependable dependency path before the video code even runs. Use Remotion when repeated video is a software product and your team already works in React.

Full review →
38

browser

lightpanda-io/browser · ★ 38k ▲1.4k/7d · Zig

Consider Lightpanda Browser for HTML, Markdown and DOM extraction when you can validate your own pages before switching. Lightpanda has no graphical rendering engine. Our run on October 1, 2026, at commit 4c748cb assessed src/rust, not a complete browser deployment. It did not test target-site compatibility, graphical output or runtime performance.

SetupDocsCommunityMaturity

Verdict Trial Lightpanda Browser for extraction when you can validate the pages and actions your application needs. In our October 1, 2026, run at commit 4c748cb, installation took 13 seconds.

Full review →
39

OpenCLI

jackwener/OpenCLI · ★ 31k ▲151/7d · JavaScript

OpenCLI is a browser automation CLI for people and AI agents. It combines reusable site adapters with controls for navigating pages, filling forms, clicking and extracting data. Its documented Browser Bridge connects Chrome or Chromium through an extension and a local daemon.

SetupDocsCommunityMaturity

Verdict OpenCLI merits a supervised trial when its site commands match a repeated task. Our measured checkout passed 7,296 tests and failed 2 in the download module; investigate those failures and the 8 audit findings before adopting that revision.

Full review →
40

kestra

kestra-io/kestra · ★ 30k ▲87/7d · Java

Kestra is a workflow orchestrator for scheduled jobs and event-driven work across data, infrastructure, and AI systems. You describe flows in YAML or edit the same definition through its visual tools, then Kestra handles triggers, retries, state, and task execution.

SetupDocsCommunityMaturity

Verdict Our Kestra run installed 11 packages in 9 seconds, but it never built or tested the 585,223-line platform because the detected npm project was only `./e2e`. Kestra deserves a trial when scheduled jobs, event triggers, scripts, and infrastructure work have become one orchestration problem.

Full review →
41

MaaAssistantArknights

MaaAssistantArknights/MaaAssistantArknights · ★ 24k ▲144/7d · C++

MAA is a desktop and command-line assistant that automates repetitive tasks in Arknights by recognizing what is on screen and controlling the game. It handles chores such as farming, base shifts, recruitment, rewards, and selected battle plans, reducing the daily attention the game demands.

SetupDocsCommunityMaturity

Verdict Our build completed in 20 seconds, but all 9 test packages in the measured Go test step failed to build. MAA is still the strongest-looking choice here for committed Arknights players because its automation breadth, platform support, integrations, and active maintenance are unusually substantial.

Full review →
42

ha_xiaomi_home

XiaoMi/ha_xiaomi_home · ★ 22k ▲29/7d · Python

Xiaomi Home is Xiaomi's official custom integration for bringing its smart-home devices into Home Assistant. It turns MIoT device properties, events, and actions into familiar Home Assistant entities, while using Xiaomi Cloud or compatible local gateway hardware to carry commands and updates.

SetupDocsCommunityMaturity

Verdict Our Xiaomi Home run installed 35 packages in 6 seconds and built in 1 second, yet pytest finished with 0 passed and 29 failed, so the official badge does not remove the need for a trial on your own Home Assistant setup. Use it when broad Xiaomi device mapping matters and Xiaomi Cloud is acceptable.

Full review →
43

moli

lexmount/moli · ★ 22k · Rust

Moli is Lexmount's open-source headless browser for AI agents. Its documentation describes JavaScript execution, DOM access, page extraction and automation through CDP, WebDriver Classic and WebDriver BiDi. Real layout requires an explicit option; our sandbox run assessed source installation and compilation, not browser compatibility.

SetupDocsCommunityMaturity

Verdict Moli built successfully in 809 seconds in our sandbox, while the test step timed out at 900 seconds with compilation still visible. The documented layout controls make it worth evaluating for extraction and DOM-based agent work, provided you can check your own pages.

Full review →
44

argo-workflows

argoproj/argo-workflows · ★ 17k ▲18/7d · Go

Argo Workflows runs container jobs on Kubernetes. Consider it when your work is already packaged as container images and your team can operate the cluster. Our stored run on 2026-10-01 at commit 98f5436 recorded unsuccessful tests and conflicting build evidence; it did not validate a deployed workflow.

SetupDocsCommunityMaturity

Verdict Consider Argo Workflows when container tasks on Kubernetes match your architecture and your team can validate the deployment. Our 2026-10-01 run at commit 98f5436 recorded unsuccessful tests and conflicting build evidence.

Full review →
45

apify-mcp-server

apify/apify-mcp-server · ★ 14k ▲816/7d · TypeScript

Apify MCP Server lets an AI assistant find and run Apify's web scrapers, crawlers, and automation jobs as tools. It gives chat and coding clients a common way to collect data from websites, search engines, maps, social networks, and stores without writing a scraper for every request.

SetupDocsCommunityMaturity

Verdict Our Apify MCP Server run installed 788 packages, then passed its 23-second build and 43-second test step. Use it when an assistant needs controlled access to approved Apify Actors and your team can govern cost, data transfer, and long jobs.

Full review →
46

runner-images

actions/runner-images · ★ 14k ▲18/7d · PowerShell

Runner Images is the source GitHub uses to assemble the virtual machines behind GitHub-hosted Actions runners and Microsoft-hosted Azure Pipelines agents. It is useful when you need to inspect what a hosted runner contains, follow image changes, or build a related Azure VM image of your own.

SetupDocsCommunityMaturity

Verdict Our sandbox could not run commit ede07f8 because its supported path is PowerShell-based and the repository has no Dockerfile, so this is a reference and image-building codebase rather than a conventional local install. Use it to audit hosted runner contents or maintain an Azure image when your team already owns the cloud plumbing.

Full review →
47

douyin-downloader

jiji262/douyin-downloader · ★ 13k ▲59/7d · Python

Douyin Downloader is a Python tool for saving Douyin videos, image posts, collections, music, comments, live streams, and whole profiles. It adds retries, local history, browser-assisted pagination, optional transcription, and a small job API for people who need more than one pasted-link download.

SetupDocsCommunityMaturity

Verdict Our Douyin Downloader run passed 649 tests in 134 seconds and pip-audit found 0 known vulnerabilities, making the Python CLI a credible choice for a permitted personal archive. Use it when cookies, local files, and occasional manual browser checks fit the job.

Full review →
48

crossplane

crossplane/crossplane · ★ 12k ▲17/7d · Go

Crossplane turns a Kubernetes cluster into an API for provisioning cloud services, applications, and other infrastructure. Platform teams define their own resource types, then Crossplane keeps the resources behind those APIs in the requested state through providers and composition functions.

SetupDocsCommunityMaturity

Verdict Our Crossplane build succeeded in 138 seconds, while the test run reported 86 passed and 3 failed of 89 before ending on the `test/e2e` package. Crossplane 2.4 is a strong choice for a staffed platform team that wants Kubernetes-native product APIs and continuous reconciliation across providers.

Full review →
49

rayburst

AnInsomniacy/motrix-next · ★ 11k ▲81/7d · TypeScript

Motrix Next is now called Rayburst, a desktop download manager for direct links, torrents, magnets, ED2K, SFTP, HLS, and DASH. It puts those jobs in one graphical queue and runs them through a bundled Aria2 Next engine instead of asking you to operate aria2 yourself.

SetupDocsCommunityMaturity

Verdict Our Motrix Next checkout installed 459 packages in 27 seconds, built in 43 seconds, and passed its tests in 34 seconds, so the source is easier to verify than the product transition is to accept. Rayburst is worth trying if you want one desktop queue for web downloads, torrents, and streamed media, and a clean start is acceptable.

Full review →
50

BrowserSkill

Tencent/BrowserSkill · ★ 9.8k ▲266/7d · TypeScript

BrowserSkill lets a shell-capable AI agent control Chrome or Edge through a local command-line tool and browser extension. It opens a separate Agent Window, can temporarily borrow a tab you approve, and reuses the login state already present in your browser.

SetupDocsCommunityMaturity

Verdict Our BrowserSkill install added 726 packages in 45 seconds, but no root build or test target was available, so the published binaries and extension deserve a separate trust review before they touch logged-in accounts. It is worth trying for supervised, low-risk browser chores where reusing your real session saves substantial setup.

Full review →
51

claude-squad

smtg-ai/claude-squad · ★ 8.7k ▲33/7d · Go

Claude Squad is a terminal interface for running several coding-agent sessions at once. It puts each Claude Code, Codex, Gemini CLI, Aider, or custom command in its own tmux session and Git worktree, then lets you inspect diffs and move finished work back into your branch.

SetupDocsCommunityMaturity

Verdict Our Claude Squad run installed 48 Go packages, built successfully, and passed all 6 tests in 91 seconds combined. It is worth using on macOS or Linux when you already know how to manage worktrees and want one screen for several agent CLIs.

Full review →
52

pi-autoresearch

davebcn87/pi-autoresearch · ★ 8.2k ▲1/7d · TypeScript

pi-autoresearch is an extension for the pi coding agent that changes code, runs a numeric measurement, keeps improvements, and reverts regressions. It stores the goal, measurement script, and experiment history in the project so the work can resume after a restart or context reset.

SetupDocsCommunityMaturity

Verdict Our pi-autoresearch run installed 127 packages in 35 seconds and passed all 46 tests, making it a low-friction trial for existing pi users with a numeric target. Its durable `.auto` record makes long sessions inspectable and resumable.

Full review →
53

skill-recorder

microsoft/skill-recorder · ★ 5.6k · TypeScript

Microsoft Skill Recorder captures a desktop task and uses GitHub Copilot to turn the recording into instructions an AI agent can reuse. It can produce a SKILL.md procedure or a scheduled automation, preferring the agent's native tools over replaying screen clicks. The intended benefit is explaining a recurring workflow by doing it once, then reviewing the resulting instructions.

SetupDocsCommunityMaturity

Verdict Our Skill Recorder run completed installation, build and tests, but npm audit reported 16 known vulnerabilities, including 14 high. It is worth a controlled trial with nonconfidential tasks if you already have Copilot access and will inspect the generated instructions and requested tools.

Full review →
54

linkedin-skills

sergebulaev/linkedin-skills · ★ 5.4k ▲205/7d · Python

LinkedIn Skills is a bundle of Claude Code and Codex instructions plus small Python clients for planning, drafting, reviewing, and publishing LinkedIn content. It can work as a copy-and-paste writing aid with no account connection, or use Apify, Publora, and Pixfaro for reading posts, publishing approved drafts, and making images.

SetupDocsCommunityMaturity

Verdict Our LinkedIn Skills install took 7 seconds, used 40 MB, and passed all 102 tests, making the bundle easy to try before connecting an account. Use it if you want editable LinkedIn writing routines and will keep a person at the approval step.

Full review →
55

mcp-server-cloudflare

cloudflare/mcp-server-cloudflare · ★ 4.4k ▲20/7d · TypeScript

Cloudflare MCP Server is a monorepo behind a set of hosted Model Context Protocol endpoints for documentation, Workers, logs, browser rendering, DNS analytics, security, and other Cloudflare products. An MCP client can inspect account data and, where a tool permits it, make changes through a narrower product-specific interface than Cloudflare's separate Code Mode server.

SetupDocsCommunityMaturity

Verdict Our Cloudflare MCP Server run installed 870 packages in 16 seconds and passed all 320 tests in 129 seconds, while the monorepo supplied no build target. Use the hosted domain server that matches one narrow job, with a token limited to that job and human approval around writes.

Full review →
56

ffmpeg-skill

kajisho5/ffmpeg-skill · ★ 4.1k ▲1.3k/7d · Python

ffmpeg-skill gives Claude Code, Cursor, Codex, and MCP clients a structured local video-editing toolkit built on FFmpeg. It packages 42 Python tools for cuts, captions, audio work, reframing, exports, and checks, while keeping media on the user's machine.

SetupDocsCommunityMaturity

Verdict Our ffmpeg-skill run installed 0 npm packages in 6 seconds, but its 71-test suite ended with 1 failure and 16 errors after FFmpeg was missing from PATH. Use it when you want guardrails around an FFmpeg installation you already manage, especially from Claude Code or an MCP client.

Full review →
57

kargo

akuity/kargo · ★ 3.8k ▲17/7d · Go

Consider Kargo if your Kubernetes team needs an explicit promotion process across environments and can operate another control plane. Kargo automates the promotion of software artifacts through lifecycle stages. In the documented quickstart, it commits updated configuration to Git, and Argo CD syncs the cluster. Our September 25, 2026 run at commit ccf601b recorded a successful build step and a failed test command. It did not establish a working cluster deployment.

SetupDocsCommunityMaturity

Verdict Kargo is worth evaluating when promotion across environments needs its own explicit process and your team can operate the supporting Kubernetes services. Our September 25, 2026 run at ccf601b recorded a successful build step but a failed test command.

Full review →
58

selenium

SeleniumHQ/selenium · ★ 35k ▲7/7d · Java

Selenium lets programs open real web browsers, click through pages, and check the results in languages including Java and Python. It gives teams a common browser-control interface for local scripts and remote browser fleets, so they can automate the same user flows across their existing tools.

SetupDocsCommunityMaturity

Verdict Our Selenium source-checkout install failed in 11 seconds because pnpm reported an ignored esbuild build script, leaving that setup unresolved. Use Selenium when its language bindings or an existing Grid match your browser coverage needs, and validate the published package separately.

Full review →
59

stagehand

browserbase/stagehand · ★ 26k ▲77/7d · TypeScript

Stagehand is a browser automation SDK for TypeScript, Python, and Go. It combines familiar page and locator methods with language-model commands that click, inspect, and extract data when fixed selectors are too brittle.

SetupDocsCommunityMaturity

Verdict Our Stagehand run installed 1,444 packages in 58 seconds and built in 41, but the 23-second test command still exited 1 after all 34 browser tests were skipped. Use it when a few change-prone browser steps justify model calls and you can keep hard assertions around the result.

Full review →
60

activepieces

activepieces/activepieces · ★ 25k ▲76/7d · TypeScript

Activepieces is a visual workflow builder for connecting apps, running scheduled jobs, and adding approval or AI steps without writing the whole integration yourself. Its TypeScript integration packages can also be exposed as MCP servers for agent clients.

SetupDocsCommunityMaturity

Verdict Our Activepieces install took 367 seconds, added 3,143 packages, and used 2,670 MB, so self-hosting it is a platform commitment rather than a small automation experiment. Use it when a visual builder, TypeScript integrations, and MCP exposure belong in the same system.

Full review →
61

SpotX

SpotX-Official/SpotX · ★ 23k ▲181/7d · PowerShell

SpotX is a PowerShell patcher for the official Spotify desktop client on Windows. It removes banner, video, and audio ads, and can also hide unwanted content sections, disable analytics, and control client updates.

SetupDocsCommunityMaturity

Verdict We could not run SpotX because our Debian sandbox does not provide its supported Windows environment, and the repo has no Dockerfile. For a technically confident Windows user, it is a focused, unusually configurable Spotify patcher with clear presets and a current codebase.

Full review →
62

whatsapp-web.js

wwebjs/whatsapp-web.js · ★ 23k ▲35/7d · JavaScript

whatsapp-web.js is a Node.js library that controls a linked WhatsApp account through WhatsApp Web in a managed browser. It lets developers send and receive messages, media, reactions, polls, and group actions without using Meta's official business API.

SetupDocsCommunityMaturity

Verdict Our run installed 427 packages in 36 seconds, then the tests failed after 9 seconds and npm audit found 20 vulnerabilities, so whatsapp-web.js needs active ownership even before WhatsApp Web changes underneath it. Use it for a personal or internal workflow when its broad Web feature access is worth running Chromium and accepting account risk.

Full review →
63

Obtainium

ImranR98/Obtainium · ★ 21k ▲185/7d · Dart

Obtainium checks Android app release pages, downloads APKs, and helps keep those apps updated without revisiting each website yourself. You choose the sources, which can include developers' repositories, F-Droid repositories, and APK hosts.

SetupDocsCommunityMaturity

Verdict Our sandbox did not run Obtainium at commit 6a17d53 because it lacked Dart support and the repository had no Dockerfile. I would shortlist it for Android users who already choose their developers' release pages and will maintain source filters.

Full review →
64

OpenWA

rmyndharis/OpenWA · ★ 16k ▲176/7d · TypeScript

Consider OpenWA for internal WhatsApp automation on a dedicated number you can afford to lose. OpenWA is an unofficial WhatsApp API gateway. It combines REST endpoints, webhooks, a dashboard and optional MCP tools around whatsapp-web.js or Baileys. Our run on October 1, 2026, at commit 09dae1a recorded successful install, build and test commands in our sandbox; it did not establish live WhatsApp delivery or account safety.

SetupDocsCommunityMaturity

Verdict OpenWA is worth evaluating for internal automation when you can accept losing or relinking a dedicated number. Our October 1, 2026, run at commit 09dae1a supports a narrower conclusion: the recorded install, build and test commands completed successfully in our sandbox.

Full review →
65

espanso

espanso/espanso · ★ 15k ▲6/7d · Rust

Espanso is a local text expander that watches for typed triggers and replaces them in desktop applications. Its YAML rules can insert plain text, dates, images, form results, or output from local scripts, giving one snippet system to Windows, macOS, and Linux users.

SetupDocsCommunityMaturity

Verdict Our Espanso checkout installed 350 packages in 41 seconds, then both build and tests failed because wxWidgets was absent. The released desktop app is still an easy recommendation for people who want local, system-wide snippets and can test their own applications.

Full review →
66

camofox-browser

jo-inc/camofox-browser · ★ 12k ▲126/7d · JavaScript

Camofox Browser wraps the Firefox-based Camoufox anti-detection engine in a REST API and MCP server for AI agents. It returns compact accessibility snapshots with stable element references, manages isolated browser sessions, and supports clicks, typing, screenshots, downloads, cookies, proxies, and structured extraction without asking an agent to parse a full page.

SetupDocsCommunityMaturity

Verdict Our Camofox Browser build passed in 7 seconds with 0 known npm vulnerabilities, but Jest stopped before executing tests because `jest-junit` was missing. Use it when an authorized agent needs compact browser controls and ordinary automation fingerprints are the blocker.

Full review →
67

Windows-MCP

CursorTouch/Windows-MCP · ★ 11k ▲461/7d · Python

Windows-MCP is a Model Context Protocol server that lets an AI client inspect and control a Windows desktop. It exposes screenshots, accessibility-tree state, mouse and keyboard input, application control, PowerShell, files, processes, clipboard data, and Registry operations.

SetupDocsCommunityMaturity

Verdict Our Windows-MCP run installed 34 packages and built in 1 second, but its tests stopped before collection because Python could not import `windows_mcp`. Use it for supervised Windows automation when accessibility-tree control is more useful than pixel-only clicking, and expose the smallest possible tool list.

Full review →
68

youtube-dl-gui

jely2002/youtube-dl-gui · ★ 9.6k ▲170/7d · Rust

Choose Open Video Downloader if you want visual controls and a desktop queue for permitted media downloads through yt-dlp. Open Video Downloader is a desktop interface for yt-dlp. It supports Windows, macOS and Linux according to its README. Our run on September 20, 2026, at commit c402ee3 recorded a successful frontend build and an unsuccessful test command. It did not verify a packaged installation or a completed media download.

SetupDocsCommunityMaturity

Verdict Evaluate Open Video Downloader if you want yt-dlp through desktop controls and can check a representative download before relying on it. Our sandbox recorded frontend build success, but its unsuccessful test command and visible missing-browser errors leave browser workflows unverified.

Full review →
69

ok-wuthering-waves

ok-oldking/ok-wuthering-waves · ★ 8.4k ▲432/7d · Python

ok-wuthering-waves is primarily a Chinese-language Windows automation app for Wuthering Waves, and a full English guide is available. It uses image recognition and simulated mouse and keyboard input to handle combat, daily tasks, and Echo farming while the game is visible, minimized, or covered.

SetupDocsCommunityMaturity

Verdict Our pytest run exited 1 after 2 seconds because `cv2` was missing, even though 11 tests and 5 subtests passed, so source contributors do not get a clean fresh-container check at commit f5f38c8. The packaged Windows app is active and unusually specific about its operating conditions, but the project's own warning makes it a poor choice for an account you cannot afford to lose.

Full review →
70

actions-runner-controller

actions/actions-runner-controller · ★ 6.6k ▲18/7d · Go

Use Actions Runner Controller if you want to operate self-hosted GitHub Actions runners through Kubernetes and can own the cluster and GitHub authentication. Actions Runner Controller scales self-hosted GitHub Actions runners on Kubernetes. Its runner scale sets respond to workflow demand; controlled runners can be ephemeral containers. Our recorded sandbox results establish command outcomes, not a working runner deployment.

SetupDocsCommunityMaturity

Verdict ARC is a reasonable choice for teams that intend to manage GitHub Actions runner capacity as a Kubernetes workload. Our October 2, 2026 run at commit 1945b42 recorded successful install and build steps followed by a failing test command.

Full review →
71

volcano

volcano-sh/volcano · ★ 6.0k ▲14/7d · Go

Consider Volcano when you already operate Kubernetes and need scheduling policies for batch workloads. Volcano is a Kubernetes-native batch scheduling system. Its documented policies include gang scheduling and fair-share, but our sandbox record does not establish that a deployed scheduler or workload worked.

SetupDocsCommunityMaturity

Verdict Volcano merits a cluster evaluation when its batch scheduling policies match your requirements. Our run recorded a successful install status, conflicting build status and diagnostics, and a failing test step.

Full review →
72

kudu

AdventDevInc/kudu · ★ 4.1k ▲45/7d · TypeScript

Kudu is a desktop and command-line system cleaner for Windows, macOS, and Linux. It scans for disposable caches and logs, manages startup and privacy settings, and includes malware checks, with deletion kept local unless optional cloud features are enabled.

SetupDocsCommunityMaturity

Verdict Our Kudu run installed 564 packages, built in 16 seconds, and passed 2,613 of 2,614 tests with 1 skip, which is strong evidence for the checked-out code. Use it first as a scan-and-review tool, especially while v2.7.0 has open reports about signed binaries being flagged and cloud permissions defaulting on.

Full review →
73

delegate-skills

amElnagdy/delegate-skills · ★ 2.5k ▲36/7d · JavaScript

delegate-skills is a collection of agent skills that lets one coding agent hand an implementation brief to a separate CLI, then bring back a structured report and working-tree diff for review. It supports direct dispatch or named fleet lanes while leaving the final commit with the orchestrator.

SetupDocsCommunityMaturity

Verdict Our harness could not run commit b19180c because JavaScript was unsupported and the repository had no Dockerfile, so we cannot vouch for its install or relay tests. delegate-skills is worth adopting when you already pay the mental cost of several agent CLIs and want their diffs to arrive in one shape.

Full review →
74

WiiUDownloader

Xpl0itU/WiiUDownloader · ★ 2.3k ▲22/7d · Go

WiiUDownloader is a desktop app that downloads Wii U games, updates, DLC, and demos directly from Nintendo's CDN without requiring title keys. It solves the problem of abandoned, Windows-only Wii U USB Helper by providing a cross-platform, open-source downloader that also decrypts files for use on a real console or the Cemu emulator.

SetupDocsCommunityMaturity

Verdict Our build succeeded in 28 seconds on a fresh Debian container and produced no test output (0 passed, 0 failed of 0), so this is a small, compiling, active tool with no automated test safety net. If you need a no-title-key Wii U downloader on macOS, Linux, or Steam Deck, yes, use it.

Full review →
75

skill-up

alibaba/skill-up · ★ 1.4k ▲55/7d · Go

skill-up is a command-line evaluator for Agent Skills, including skills used by Claude Code, Codex, Qoder CLI, and Qwen Code. It runs YAML-defined cases, grades the results, produces local or CI reports, and ships a companion skill that can revise a skill or its eval suite after failures.

SetupDocsCommunityMaturity

Verdict Our npm run installed 150 packages in 35 seconds, but the package exposed no build or test target, so it did not verify the Go evaluator. skill-up is worth adopting when Agent Skills already affect real work and failures need to become repeatable regression cases across several clients.

Full review →
76

agent-deck

asheshgoplani/agent-deck · ★ 1.2k ▲31/7d · Go

Choose Agent Deck if switching between coding-agent sessions and preparing separate worktrees has become a recurring chore. Start with supervised local sessions: our run on October 1, 2026, at commit 035fd60 did not establish that a working agent fleet could run unattended. Agent Deck is a terminal session manager for AI coding agents. Its documented features include session status, grouping, search, forking, Git worktrees, MCP management and a remote conductor.

SetupDocsCommunityMaturity

Verdict Evaluate Agent Deck for supervised management of multiple coding-agent sessions. Our October 1, 2026 run at commit 035fd60 completed installation and build, but tests timed out and the raw tail exposed missing tmux.

Full review →
77

herdr-projects

eliasstravik/herdr-projects · ★ 1.0k ▲148/7d · Rust

Herdr Projects is a Herdr plugin that gives one coordinator agent a set of parallel worker threads, each with its own branch or folder, shared instructions, and project memory. It tracks which threads are working, blocked, or ready for review so you do not have to brief and watch every coding agent by hand.

SetupDocsCommunityMaturity

Verdict Our v0.2.34 run built in 44 seconds and passed 648 of 650 tests, with one logged failure caused by a missing `rsync` executable, so the core is substantial but the fresh-container suite is not clean. Herdr Projects is worth trying if Herdr already runs your agents and your main problem is coordination across branches and machines.

Full review →
78

claude-ads

AgriciDaniel/claude-ads · ★ 11k ▲195/7d · Python

Consider Claude Ads for supervised paid-media audits if your team already uses Claude Code and can check findings against the original account data. Claude Ads is an open-source Claude Code skill with Python tooling for audits, plans, creative workflows, monitoring and reports across twelve advertising platforms. The current README labels Claude Ads a development candidate and says all twelve account-health scoring profiles are disabled.

SetupDocsCommunityMaturity

Verdict Claude Ads is worth evaluating for supervised audits by teams already using Claude Code. Our run installed and built successfully, but the test command exited 1 after a missing-jsonschema collection error, and pip-audit recorded 7 advisories with unknown scan coverage.

Full review →
79

hey-cli

basecamp/hey-cli · ★ 422 ▲4/7d · Go

Choose HEY CLI if you use HEY and want terminal access to its mail and calendar alongside coding-agent tools. It provides a CLI and TUI for HEY; it is a companion to the service. Our run on 2026-10-10 at commit 732568e recorded successful install, build and test commands, but did not establish that signed-in mail, calendar or agent workflows work.

SetupDocsCommunityMaturity

Verdict HEY CLI is worth evaluating for an existing HEY customer who wants terminal and coding-agent access. Our run on 2026-10-10 at commit 732568e supports source installation, build and test-command success in our sandbox.

Full review →
80

paperclip

paperclipai/paperclip · ★ 109k ▲1.7k/7d · TypeScript

Paperclip is a self-hosted control plane for assigning work to teams of AI agents and tracking their tasks, costs, permissions, approvals, and history. It connects existing workers such as Claude Code, Codex, OpenClaw, shell processes, and HTTP agents instead of supplying a new agent model.

SetupDocsCommunityMaturity

Verdict Our Paperclip build stopped after 50 seconds because `cargo` was unavailable, and its tests ran into our 900-second cap, so this is not a Node-only install you can judge from a quick dashboard launch. Use it when many agents already create real ownership, budget, and audit problems.

Full review →
81

openpilot

commaai/openpilot · ★ 64k ▲39/7d · Python

Openpilot connects comma hardware to a supported car to assist with driving. It gives owners and developers access to the software behind that assistance, while requiring an attentive driver who can take control.

SetupDocsCommunityMaturity

Verdict Our openpilot run installed 63 packages in 27 seconds and built successfully in 8 seconds, but skipped tests because no runnable target was found. Consider it for a supported car with the documented hardware and an attentive driver.

Full review →
82

spiderfoot

smicallef/spiderfoot · ★ 24k ▲301/7d · Python

Choose SpiderFoot when an authorized investigation needs collection across domains, IP addresses, accounts, and other identifiers, and an analyst can verify the findings. SpiderFoot automates OSINT through a web interface or CLI. Our run on October 2, 2026, used commit 0f815a2. It recorded an unfinished test command and dependency advisories, without establishing scan accuracy or deployment readiness.

SetupDocsCommunityMaturity

Verdict Evaluate SpiderFoot for authorized investigations that need collection across several identifier types and allow analyst verification. Our stored run supports a setup and dependency assessment, but it does not verify an operational scan or deployment.

Full review →
83

jev-ultrafast

browser-use/jev-ultrafast · ★ 24k ▲297/7d · Python

Jev Ultrafast is a Python browser agent that chooses from visible page controls instead of asking a language model to invent selectors or scripts. It is built for short, bounded web tasks where fewer model round trips matter, with a second small model supplying text only when the agent selects a field.

SetupDocsCommunityMaturity

Verdict Our Jev Ultrafast run installed 48 packages in 19 seconds and passed all 31 tests, so the code is easy to evaluate even though the browser coverage is still MVP-sized. Use it for bounded experiments where every outcome has an independent check and the supported DOM controls match the site.

Full review →
84

Archon

coleam00/Archon · ★ 24k ▲23/7d · TypeScript

Archon is an open-source workflow engine that turns an AI coding process into a repeatable YAML-defined sequence. It coordinates planning, implementation, testing, review, approval, and pull-request creation so an agent is less likely to skip a step or improvise the process each time.

SetupDocsCommunityMaturity

Verdict Our build finished in 59 seconds, but the full test command ended with 26 failures after 353 seconds. Archon has a sharp, useful idea: keep AI judgment inside a process that your team owns and can review.

Full review →
85

TikTokDownloader

JoeanAmier/TikTokDownloader · ★ 17k ▲136/7d · JavaScript

DouK-Downloader is Chinese-first software, and a full English README is available. It downloads and records Douyin or TikTok posts, account feeds, images, and live streams, while also saving collected metadata to files or exposing it through a local API.

SetupDocsCommunityMaturity

Verdict Our DouK-Downloader run installed 57 packages in 31 seconds and built in 8 seconds, but it offered no test target and cannot promise maintained request signing. Use version 5.8 for supervised, authorized archives when you can refresh Cookies, inspect output, and repair workflows after platform changes.

Full review →
86

midscene

web-infra-dev/midscene · ★ 15k ▲59/7d · TypeScript

Midscene is a TypeScript GUI automation system that asks a multimodal model to find and operate controls from screenshots. It plugs into Playwright, Puppeteer, Vitest, YAML workflows, and agent skills, with adapters extending the same approach to mobile and desktop interfaces.

SetupDocsCommunityMaturity

Verdict Our Midscene checkout built in 13 seconds, but its 2,236-package install used 2,057 MB and the full test target exited 1 despite 88 reported passes and 0 failed cases. Trial it for visual surfaces that defeat selectors, especially canvas, cross-origin, and native interfaces.

Full review →
87

OpenMausBot

milind-soni/OpenMausBot · ★ 8.6k ▲3.0k/7d · TypeScript

OpenMausBot is a local-first desktop chat app in English where each contact is a separate Claude, Codex, or Grok agent running through a CLI on your computer. It gives those agents their own conversations, models, approvals, optional cloud or local computers, connected apps, routines, and voice features.

SetupDocsCommunityMaturity

Verdict Our OpenMausBot run built in 21 seconds and passed 1,692 of 1,701 tests, but one process-cleanup test failed after the 1,405 MB install, so the current source is substantial and still has a reliability edge to investigate. Try it if a local desktop roster is more useful than one agent terminal and you can keep approvals enabled.

Full review →
88

flux2

fluxcd/flux2 · ★ 8.5k ▲25/7d · Go

Flux version 2 keeps Kubernetes clusters in sync with configuration sources such as Git repositories and OCI artifacts. Its GitOps Toolkit supplies Kubernetes custom resources and controllers for sources, Kustomize, Helm, notifications and image automation. These capabilities are documented by the project; our recorded run did not exercise cluster reconciliation.

SetupDocsCommunityMaturity

Verdict Flux is worth evaluating for Kubernetes teams that want Git-managed configuration and composable controllers. Our 2026-10-01 run at commit 3fd91f5 did not verify a working CLI or cluster deployment: the build-labelled step contains an embedded-manifest error despite its zero exit status, and the test command failed during cmd/flux setup.

Full review →
89

flyte

flyteorg/flyte · ★ 7.8k ▲74/7d · Go

Evaluate Flyte 2 for Python ML workflows if your team can operate Kubernetes and PostgreSQL; defer a self-hosting commitment until you verify the backend delivery path. The Flyte README says its open source backend is coming soon. Flyte 2 provides a Python interface for pipelines, models and agents, while this repository contains backend infrastructure and API definitions. The full Python SDK lives in a separate repository.

SetupDocsCommunityMaturity

Verdict Flyte 2 merits a scoped evaluation by teams prepared to operate Kubernetes and PostgreSQL. Verify the backend delivery path before choosing it for self-hosting.

Full review →
90

md2wechat-skill

geekjourneyx/md2wechat-skill · ★ 3.8k ▲14/7d · Go

md2wechat is a Chinese-first CLI and agent skill that turns Markdown into formatted WeChat Official Account articles, checks them, creates images, and can send finished work to the WeChat draft box. The README and linked operating guides are in Chinese; the license contains an English legal text, but the repository does not list a full English user guide.

SetupDocsCommunityMaturity

Verdict Our md2wechat run installed 0 packages in 13 seconds, but the repository exposed no build or test command to verify the shipped CLI path. It is a sensible fit for Chinese WeChat creators who want agent-readable checks and accept the paid renderer, credential work, and source-available license.

Full review →
91

karpenter

kubernetes-sigs/karpenter · ★ 2.3k ▲13/7d · Go

Evaluate Karpenter when pending pods need additional nodes and your team can validate the chosen provider implementation. Karpenter provisions nodes that meet pod requirements and removes nodes when they are no longer needed. Our run on October 1, 2026, at commit 7327802 recorded successful install and build steps, followed by a failed test command. It did not validate a cloud deployment or production workload.

SetupDocsCommunityMaturity

Verdict Evaluate Karpenter if pod requirements should drive node provisioning and your team can validate the provider separately. Our October 1, 2026 sandbox run at commit 7327802 completed install and build, but the test command failed.

Full review →
92

agent-desktop

lahfir/agent-desktop · ★ 1.8k ▲27/7d · Rust

Agent Desktop lets an AI agent inspect and control macOS applications through the operating system's accessibility tree, returning JSON and reusable element references instead of depending on screenshots alone. It covers observation, clicks, typing, windows, clipboard, notifications, sessions, and traces; Windows and Linux desktop control are still listed as planned.

SetupDocsCommunityMaturity

Verdict Our Debian run installed 65 packages in 22 seconds, then Agent Desktop failed both build and test compilation, so Linux buyers should not read its multi-platform library artifacts as working desktop control. On macOS 13+, its structured refs, action checks, and trace format make it a serious candidate for an agent that must operate native apps.

Full review →
93

gawkbot

najmuzzaman-mohammad/gawkbot · ★ 1.5k ▲10/7d · Go

gawkbot turns a described or demonstrated work routine into a locally run bot and a small app for managing its results. It is meant to automate recurring work across more than 1,200 integrations while holding sends, commits, purchases, and deletions for human approval.

SetupDocsCommunityMaturity

Verdict Our sandbox install succeeded in 151 seconds, but this checkout exposed no build or test target for us to run. gawkbot is worth a careful local trial if its approval-first approach solves a real workflow and you already use one of its supported coding runtimes.

Full review →
94

ongrid

ongridio/ongrid · ★ 1.2k ▲22/7d · Go

Ongrid is a self-hosted operations workspace where an AI agent can inspect metrics, logs, traces, topology, runbooks, and hosts from a web console or team chat. It aims to shorten incident diagnosis, then puts production-changing commands behind a human approval flow.

SetupDocsCommunityMaturity

Verdict Our Ongrid run built in 104 seconds and passed all 117 discovered tests, which supports a serious read-only pilot for a Linux fleet. Do not enable production write actions until your exact release proves that redirection, pipes, substitutions, and compound commands all enter the approval flow.

Full review →
95

warp-masque-actions

byJoey/warp-masque-actions · ★ 1.1k ▲9/7d · JavaScript

Warp Masque Actions is a Chinese-language set of GitHub Actions and a Cloudflare Worker that generate Cloudflare WARP MASQUE credentials and mihomo configurations. The documentation is detailed, but there is no English README, so non-Chinese readers will need translation before handling its credential and client setup.

SetupDocsCommunityMaturity

Verdict Our worker run installed 45 packages, used 231 MB, and passed its build and tests in 37 seconds combined, so the code path is small enough to inspect and reproduce. Use Warp Masque Actions if you read Chinese, already accept mihomo Alpha, and understand that 57 endpoints still mean one WARP exit.

Full review →
96

link-cli

stripe/link-cli · ★ 884 ▲5/7d · TypeScript

Link CLI gives software agents one-time payment credentials from a user's Link wallet, so an agent can complete a purchase without storing the user's real card details. It supports ordinary checkout forms, Stripe-hosted payment pages, and sellers using machine payment protocols, with the user approving a spend request before credentials are issued.

SetupDocsCommunityMaturity

Verdict Our build finished in 7 seconds, and all 166 tests passed, which makes Link CLI unusually credible for a project still at version 0.14.0. Use it if you are building a US-focused purchasing agent and want approval-gated credentials without exposing a real card.

Full review →
97

foreman

thruwire/foreman · ★ 799 ▲13/7d · Python

Try Foreman for bounded repository jobs when you want a separate supervisor around an existing coding agent and can inspect its interventions. Foreman supervises one coding worker at a time. It assesses progress, completion, verification needs and human intervention; the worker still chooses tools and edits files. Our run on October 4, 2026 tested commit e5d1aa4 in our sandbox, not authenticated coding-agent supervision or assessment accuracy.

SetupDocsCommunityMaturity

Verdict Use Foreman for an inspected trial of single-worker supervision, starting with the documented deterministic demo and then a bounded repository job. Our October 4, 2026 run at e5d1aa4 established successful installation and build alongside a failing test command.

Full review →
98

stop-stutter

burakgon/stop-stutter · ★ 451 ▲2/7d · Swift

Stop Stutter is a native macOS utility that keeps Apple's AWDL peer-to-peer interface off while selected game-streaming clients are running. It automates one specific Wi-Fi workaround; it does not fix stutter caused by router congestion, packet loss, decoding, or display timing.

SetupDocsCommunityMaturity

Verdict Our harness installed 35 packages and built in 27 seconds combined, but it skipped tests and did not exercise the Swift app or privileged helper. Stop Stutter is worth trying only after an off-versus-on check shows that AWDL causes your streaming hitches.

Full review →
99

qiaomu-download

joeseesun/qiaomu-download · ★ 435 ▲8/7d · Python

qiaomu-download is a Chinese-language Agent Skill for downloading video, audio, or subtitles through a plain request such as "download this URL." The main documentation is in Chinese and there is no full English README; the skill wraps yt-dlp, adds post-download ffprobe checks, and has special paths for WeChat Channels and Spotify tracks.

SetupDocsCommunityMaturity

Verdict Our qiaomu-download run installed 35 packages in 11 seconds and passed all 43 tests, so its small Python layer is easy to verify. Use it if you read Chinese and want an agent to apply strict download, cookie, and ffprobe rules around yt-dlp.

Full review →
100

k-skill

NomaDamas/k-skill · ★ 8.2k ▲99/7d · JavaScript

Choose k-skill if you need Korean service helpers for a coding agent and can check each skill's credentials and action limits. The collection supports Claude Code, Codex, OpenCode and other agents. Our run on October 2, 2026, at commit 9fade13 completed installation and build, but the test command returned exit 1. We did not verify live service workflows, authenticated actions or deployment.

SetupDocsCommunityMaturity

Verdict Start with one Korean service task, inspect its skill instructions and credential route, and verify the output against the underlying service. Our sandbox recorded successful installation and build at commit 9fade13, a Python test-preparation failure and three high-severity npm advisories.

Full review →
101

go-whatsapp-web-multidevice

aldinokemal/go-whatsapp-web-multidevice · ★ 5.0k ▲24/7d · Go

Consider GOWA for a supervised trial of a self-hosted WhatsApp HTTP API with MCP access, but resolve its failing sandbox test command before adopting it. GOWA can manage multiple WhatsApp accounts in one server instance. In our run on October 2, 2026, at commit a8351fb, install and build succeeded, while the test command exited with code 1. This record does not establish successful phone pairing, live messaging, Docker deployment, or MCP client operation.

SetupDocsCommunityMaturity

Verdict Evaluate GOWA with a dedicated test account after reproducing the failing test command and collecting complete logs. Our run supports successful install and build steps, not successful live operation.

Full review →
102

MoneyPrinterTurbo

harry0703/MoneyPrinterTurbo · ★ 137k ▲1.5k/7d · Python

MoneyPrinterTurbo is a Python application for assembling narrated short videos from a topic or supplied script. Its README describes WebUI, API, CLI and AI Agent workflows covering scripts, footage, voiceover, subtitles, music, rendering and optional social publishing. The default README is Chinese; an English README and a Japanese documentation link are available. Our recorded run checked dependency installation, a build step and tests, not finished-video quality.

SetupDocsCommunityMaturity

Verdict MoneyPrinterTurbo is worth a controlled trial if its documented script-to-video workflow fits your publishing process. Our September 27, 2026 run at ad5496f completed installation, build and tests, but did not demonstrate video rendering or an installed CLI.

Full review →
103

gkd

gkd-kit/gkd · ★ 44k ▲380/7d · Kotlin

GKD is an Android app that clicks screen elements when your chosen rules match, helping with repeated confirmations and unwanted startup screens. Its README and official guides are primarily Chinese, and I found no official English documentation in the linked material.

SetupDocsCommunityMaturity

Verdict Our GKD run installed 6 packages in 5 seconds and used 35 MB, but skipped build and tests, leaving Android behavior unverified. Use it for specific screen actions you can inspect and maintain through rules.

Full review →
104

invisible_dots

feder-cr/invisible_dots · ★ 32k ▲69/7d · TypeScript

Invisible Dots runs each AI agent inside its own persistent Linux virtual machine, with a desktop, shell, files, scheduled jobs, and a patched Firefox browser. You control the agents from a local web UI, CLI, API, Telegram, or an optional unofficial WhatsApp connection, while OpenRouter supplies the model.

SetupDocsCommunityMaturity

Verdict Our Invisible Dots run built in 40 seconds and passed 2,102 of 2,106 Vitest cases, but one secret-input test failed, so its broad suite was close rather than clean. Try it if you specifically want persistent, permissioned VMs for browser agents and can spare the hardware.

Full review →
105

cli

googleworkspace/cli · ★ 32k ▲125/7d · Rust

Google Workspace CLI, invoked as gws, lets scripts and assistants read and change Google services such as Drive, Gmail, Calendar, and Sheets. It turns Google's API descriptions into terminal commands with JSON output, saving you from writing a separate HTTP wrapper for each service.

SetupDocsCommunityMaturity

Verdict Our Google Workspace CLI run installed 104 packages in 7 seconds and used 36 MB, then its test command stopped because Cargo was missing. Use it for inspectable Workspace scripts if you can handle OAuth setup and pin a version against the documented breaking changes.

Full review →
106

GitHub520

521xueweihan/GitHub520 · ★ 30k ▲38/7d · Python

GitHub520 publishes a frequently updated hosts file that maps GitHub domains to selected IP addresses. Its documentation is written in Chinese, with no English README, and it is aimed mainly at people whose GitHub pages or images load slowly or fail to load.

SetupDocsCommunityMaturity

Verdict Our GitHub520 run installed 61 packages in 30 seconds and pip-audit found 12 known vulnerabilities, even though ordinary users do not need the Python updater. Use the published hosts feed as a reversible personal workaround when GitHub access is unreliable and you can inspect each update.

Full review →
107

node-red

node-red/node-red · ★ 24k ▲15/7d · JavaScript

Node-RED is a browser-based tool for building event-driven applications with visual flows. It suits teams connecting devices and services that can also take responsibility for editor access, stored state, and deployment.

SetupDocsCommunityMaturity

Verdict Choose Node-RED when visual event flows fit your application and your team can own authentication, persistence, and deployment checks. Our October 8, 2026 run at d8d13d6 completed installation, build, and tests successfully; the log reported 2,878 passing and 60 pending, while npm audit reported 27 advisories with unknown coverage.

Full review →
108

suna

kortix-ai/suna · ★ 20k ▲23/7d · TypeScript

Kortix, in the repository still named Suna, is a system for running AI agents inside isolated cloud computers and reviewing their work through git-style change requests. It gives a company one place to version agent instructions, skills, memory, connectors, schedules, and the machine configuration behind automated work.

SetupDocsCommunityMaturity

Verdict Our Suna install pulled 2,651 packages and 3,393 MB, then build and tests both failed within 4 seconds because Bun was missing. Kortix is worth evaluating when isolated agent computers, reviewed change requests, and shared company configuration are the actual requirements.

Full review →
109

goreleaser

goreleaser/goreleaser · ★ 16k ▲3/7d · Go

GoReleaser is a release automation tool configured through .goreleaser.yaml. Its documentation describes building, archiving, packaging, signing, and publishing artifacts for Go, Rust, Zig, JavaScript, TypeScript, and Python projects. Which steps run depends on your configuration and edition.

SetupDocsCommunityMaturity

Verdict GoReleaser is worth evaluating when artifact building, packaging, and publication need one maintained configuration. Check Pro requirements before designing the pipeline.

Full review →
110

artemis

google/artemis · ★ 16k ▲1.0k/7d · Python

Artemis lets a coding assistant or a plain-language request operate Android apps on a connected phone or emulator. It collects screenshots, device logs and execution traces so developers can investigate what happened during an automated journey.

SetupDocsCommunityMaturity

Verdict Our Artemis run installed 208 packages and built successfully, then tests stopped after 8 seconds because OpenCV could not load libGL.so.1. Try it for supervised Android exploration and diagnostic traces if you can maintain the device host and model configuration.

Full review →
111

automatisch

automatisch/automatisch · ★ 14k ▲5/7d · JavaScript

Automatisch is a self-hosted workflow builder for connecting online services and moving data between them without writing each integration from scratch. It targets the same trigger-and-action work as Zapier, while keeping workflow data and third-party credentials on infrastructure you control.

SetupDocsCommunityMaturity

Verdict Our Automatisch install pulled 727 packages in 71 seconds, but its tests stopped after 8 seconds because the required `.env.test` file was missing. It is worth a controlled trial for a team that wants visual automation on its own servers and can own four services, durable secrets, and integration credentials.

Full review →
112

microduck

pollen-robotics/microduck · ★ 13k ▲1.5k/7d · Rust

Microduck is the operating software for Pollen Robotics' small biped robot, covering motor control, device configuration and camera services. It loads learned movement policies and manages software updates; training those policies happens in a separate repository.

SetupDocsCommunityMaturity

Verdict Our Microduck run installed 547 packages, then its build and test commands both failed because pkg-config could not find GStreamer. Use this repository to work on Microduck's runtime if you can maintain the Linux libraries and board configuration its documentation requires.

Full review →
113

escrcpy

viarotel-org/escrcpy · ★ 12k ▲69/7d · JavaScript

Escrcpy is a desktop app for viewing and controlling Android devices through a graphical interface built around scrcpy. It brings mirroring, keyboard mapping, multi-device operations, wireless connections, automation, and an MCP-based assistant into one place, reducing the need to juggle separate ADB commands and utilities.

SetupDocsCommunityMaturity

Verdict Our clean-container build failed after 24 seconds with exit code 127. Escrcpy is appealing for people who have outgrown plain scrcpy and genuinely need multi-device control, keyboard mapping, or visual automation in one desktop app.

Full review →
114

Ghost-Downloader-3

XiaoYouChR/Ghost-Downloader-3 · ★ 11k ▲239/7d · Python

Ghost Downloader is a desktop download manager from a Chinese-speaking project community, with an English README and English user documentation. It combines file downloads, torrents, and web-video handling, and its browser extension hands downloads to the desktop app so you can manage them in one queue.

SetupDocsCommunityMaturity

Verdict Our Ghost Downloader run installed 34 packages in 18 seconds, but tests stopped after 1 second because aiohttp was missing. Try the packaged desktop app if browser capture, torrents, and web video belong in your daily queue, while treating source setup and integration checks as separate work.

Full review →
115

web-access

eze-is/web-access · ★ 9.3k ▲43/7d · JavaScript

Web Access is a Chinese-first Agent Skill that teaches Claude Code and other SKILL.md-capable agents how to choose web tools and drive a logged-in Chrome or Edge session through CDP. An English translation exists through the README language menu, but the repository documentation and issue discussion are primarily Chinese.

SetupDocsCommunityMaturity

Verdict Our sandbox produced no install, build, or test result for Web Access because its JavaScript skill layout was unsupported and it ships no Dockerfile. Version 2.5.4 is worth reading if you use Claude Code in Chinese and explicitly want an agent inside your current Chrome or Edge session.

Full review →
116

pipeline

tektoncd/pipeline · ★ 9.1k ▲3/7d · Go

Tekton Pipelines provides Kubernetes resources for declaring CI/CD pipelines built from containers. It suits teams that want pipeline execution inside their Kubernetes platform and can operate the installation themselves.

SetupDocsCommunityMaturity

Verdict Shortlist Tekton Pipelines when your platform team wants CI/CD expressed as Kubernetes resources and can own installation and upgrades. Our October 1, 2026 run at commit 61e0221 recorded a successful go test command and 151 successful package results, but it did not verify a Kubernetes deployment or production workload.

Full review →
117

DLSS5-Swapper

rakanki911/DLSS5-Swapper · ★ 8.5k ▲366/7d · JavaScript

DLSS 5 Swapper is a Windows desktop app that finds installed games, adds one of several neural-rendering routes, and keeps backups for restoration. It covers native DLSS games, older graphics APIs, and emulators, with per-game controls and community compatibility reports.

SetupDocsCommunityMaturity

Verdict Our source run installed 323 packages, but the build stopped for a missing DLSS 5 payload and 6 of 278 tests failed, so this is not a clean build-from-checkout project. The packaged Windows app still earns a cautious trial for an RTX owner who will test one game at a time and use Restore originals before changing routes.

Full review →
118

loopx

huangruiteng/loopx · ★ 6.5k ▲66/7d · Python

LoopX is a local-first control plane for keeping agent goals, decisions, ownership and evidence across sessions and runtimes. Your agent host supplies the model and tools. The current README presents LoopX 2.0 as a shared TypeScript control plane and research workbench; our recorded dependency-install run predates that release and does not validate its operation.

SetupDocsCommunityMaturity

Verdict LoopX is worth a bounded trial if your agent work needs durable goals, explicit ownership and reviewable handoffs across sessions. Our 2 October 2026 run at commit 48895f7 established a successful npm dependency-install step, with build and tests skipped.

Full review →
119

user-scanner

kaifcodec/user-scanner · ★ 5.5k ▲52/7d · Python

User Scanner is a Python OSINT tool for finding possible accounts from usernames and email addresses. Its README documents profile metadata extraction, cross-scanning between identifiers, JSON, CSV and PDF reports, a Python interface and an optional MCP server. These capabilities produce investigative leads; they do not establish identity.

SetupDocsCommunityMaturity

Verdict User Scanner is worth evaluating for investigator-led username and email research where someone will review profile links, collisions and inconclusive responses. Our October 4, 2026 run at commit 780d077 recorded successful dependency-install and build steps, plus a pytest summary of 386 passed, 3 skipped and 1 collection error caused by a missing mcp import.

Full review →
120

yichen-skills

mcncarl/yichen-skills · ★ 5.1k ▲308/7d · Python

Yichen Skills is a collection of 21 Claude Code and Codex workflows for writing, research, social publishing, transcription, and authorized local WeChat or WeCom data work. The root documentation is available in English and Chinese, although individual skill instructions vary. Each directory is its own tool with separate platform, account, privacy, and licensing requirements.

SetupDocsCommunityMaturity

Verdict Our run of the X uploader installed 39 packages and passed all 89 tests with 0 known vulnerabilities, but that evidence covers one subproject inside a 21-skill collection. Adopt Yichen Skills one directory at a time after checking its platform, data access, and license.

Full review →
121

agenta

Agenta-AI/agenta · ★ 4.9k ▲16/7d · TypeScript

Agenta is a shared workspace for building agents through chat, connecting them to tools, and running them interactively or on schedules and app events. It adds files, team access, configuration history, traces, costs, permissions, and background execution around harnesses such as Claude Code, Pi, and Codex.

SetupDocsCommunityMaturity

Verdict Our Agenta API install and build finished in 35 seconds combined, but pytest stopped after 6 seconds because two configured arguments were unrecognized. Trial Agenta when you need Claude Code or Codex to live inside a shared, traceable workspace and you can operate its multi-service stack.

Full review →
122

upload-artifact

actions/upload-artifact · ★ 4.2k ▲3/7d · TypeScript

GitHub's actions/upload-artifact stores workflow output as downloadable artifacts. Its README documents file, directory and wildcard uploads, configurable retention and compression, and outputs for an artifact ID, download URL and SHA-256 digest. Downloads require a signed-in user and remain available only until expiry or deletion.

SetupDocsCommunityMaturity

Verdict Use upload-artifact v7 when signed-in, expiring workflow downloads fit the job and the README's platform restrictions are acceptable. Our October 11, 2026 run at cf430e0 completed dependency installation, build and tests; Jest reported 31 passing tests.

Full review →
123

background-agents

ColeMurray/background-agents · ★ 3.4k ▲15/7d · TypeScript

Background Agents, branded Open-Inspect in its documentation, is a self-hosted system for sending coding work to remote sandboxes and checking the result later. It connects a web app, Slack, GitHub, Linear, schedules, and webhooks to coding agents that can edit repositories and open pull requests.

SetupDocsCommunityMaturity

Verdict Our Background Agents run installed 1,296 packages and 1,184 MB, built in 64 seconds, then exited 1 after a 408-second test run whose log included 18 failures and 1 error. It is worth studying or piloting for a trusted engineering organization that wants shared background coding sessions and can operate the cloud pieces.

Full review →
124

Ydisks-Xianyu-Helper

Christ9038/Ydisks-Xianyu-Helper · ★ 2.2k ▲78/7d · Go

Ydisks Xianyu Helper is a self-hosted console for Xianyu seller accounts, buyer messages, and automated delivery of digital goods. Chinese is the primary community language, with an English README and operational wiki available. It brings order rules and inventories of codes, links, or images into the same interface.

SetupDocsCommunityMaturity

Verdict Our Ydisks run passed all 104 tests after a 20-second install and 55-second build, making the codebase a credible candidate for a supervised Xianyu seller trial. Use it when order rules and digital-goods inventory justify maintaining account credentials and a database.

Full review →
125

newsjack

elvisun/newsjack · ★ 1.7k ▲12/7d · Go

Newsjack is an English-language bundle of 30 PR skills plus a Go command-line tool for AI agents. It helps founders and communications teams monitor news, judge whether they have a credible angle, check claims, find journalists, and draft material without assembling each workflow from scratch.

SetupDocsCommunityMaturity

Verdict Our Newsjack install took 1 second and only 13 MB, but the repository exposed no build or test target to our npm harness, so adoption rests more on workflow fit than verified package behavior. Use it if your team already works in a local coding agent and wants a firm PR process around that habit.

Full review →
126

typesafe-computer-use

awlevin/typesafe-computer-use · ★ 1.3k ▲12/7d · Python

TypeSafe Computer Use, also called jev, is a desktop agent that turns a screen into a bounded list of actions and asks a small classifier what to do next. Code supplies facts such as dates, focus, and URLs, while a separate writing model is used only for free text and final answers.

SetupDocsCommunityMaturity

Verdict Our run passed 729 tests with 0 failures after an 18-second install, so TypeSafe Computer Use has far more engineering underneath it than most two-week-old agent experiments. Use it to study bounded-action computer control or to prototype supervised Mac tasks where cheap decisions justify writing deterministic state.

Full review →
127

MacDuo

DhananjayBhosale/MacDuo · ★ 230 · Swift

Mac Duo is a native macOS utility that bends, rolls, shutters, or fades your desktop as you move a MacBook lid. It reads the machine's lid-angle sensor and renders six local Metal effects, giving the physical screen movement a matching desktop animation.

SetupDocsCommunityMaturity

Verdict Mac Duo v0.1.15 delivers six lid-tracked effects, but its public builds are not notarized and several common MacBook models are unsupported. Try it on a compatible personal Mac if the effect delights you and the Screen Recording permission is acceptable.

Full review →
128

fullsend

fullsend-ai/fullsend · ★ 166 ▲4/7d · Go

Fullsend v0.43.0 connects separate coding agents to GitHub, GitLab, and Forgejo so they can triage issues, write fixes, review pull requests, and merge approved work. It is an automation system for whole repositories, with dedicated identities and policy controls, rather than a chat assistant you run for one coding session.

SetupDocsCommunityMaturity

Verdict Our run installed 624 npm packages in 27 seconds and passed all 75 Vitest tests, but that clean result covers Fullsend's web workspace rather than an autonomous production fleet. Use Fullsend if your organization is ready to treat coding agents as infrastructure, with separate identities, guarded paths, spend controls, and shadow-mode evidence.

Full review →
129

krops

polarsquad/krops · ★ 165 · Rust

Krops is an English-language reference implementation for managing cloud infrastructure through Kubernetes resources and GitOps. It combines Flux, Cluster API, and existing cloud controllers instead of adding its own controller or resource type. Platform teams can study or fork it to manage clusters and AWS resources from the same YAML, policy, and audit system used for workloads.

SetupDocsCommunityMaturity

Verdict Our krops build took 75 seconds and all 100 Rust tests passed in 44 seconds, making the bootstrap code the strongest measured result in this group. Fork it when your platform team already trusts Flux and Cluster API, and when AWS plus local Docker cover the environments you can verify yourself.

Full review →
130

agent-fleet-manager

dreamers-laboratory/agent-fleet-manager · ★ 139 · Python

Agent Fleet Manager is a small Python engine that keeps a schedule of sources to check, hands due work to workers, and records the results in one SQLite database. It solves the bookkeeping around repeated fetches or agent jobs, including retries, change detection, and an audit trail, but you supply any worker that needs more than a plain HTTP request.

SetupDocsCommunityMaturity

Verdict Our run installed 35 packages in 7 seconds, built in 1 second, and passed all 7 tests in 2 seconds, so Agent Fleet Manager is cheap to try but too young to trust as a distributed control plane. Use it when one trusted Python process needs clear scheduling and receipt bookkeeping for a modest fleet.

Full review →
131

jevry

michaelswissa/jevry · ★ 118 ▲1/7d · TypeScript

Jevry is a desktop browser that follows plain-language tasks, works through supported website controls, and returns research with source links. It keeps the browser actions visible so you can check progress, stop the task, or redirect it.

SetupDocsCommunityMaturity

Verdict Our Jevry run built in 13 seconds and passed 603 tests with 101 skipped, so the source preview earns a developer trial without proving live browsing reliability. Use it for visible research and supported page tasks when you can supply both model connections and check the outcome.

Full review →
132

omniget

tonhowtf/omniget · ★ 113 ▲294/7d · Rust

OmniGet is a desktop app for downloading online video, paid courses you can already access, galleries, torrents, and direct files. It wraps yt-dlp, FFmpeg, gallery-dl, and native extractors in one queue, then adds playback, reading, conversion, and study tools.

SetupDocsCommunityMaturity

Verdict Our OmniGet run installed 367 packages in 16 seconds, passed a 71-second build and a 6-second test step, but npm audit found 9 known vulnerabilities. Use v0.11.0 when one graphical home for yt-dlp, course files, local media, and agent tools is worth that broad trust boundary.

Full review →
133

acme.sh

acmesh-official/acme.sh · ★ 48k ▲34/7d · Shell

acme.sh is an ACME certificate client written in Unix Shell. Its README documents certificate issuance, renewal and installation without a Python dependency in the core client. Unattended operation depends on the validation method and a working deployment or reload step.

SetupDocsCommunityMaturity

Verdict acme.sh is worth evaluating when you want a Shell ACME client and can own validation, certificate placement and service reloads. Our October 9, 2026 run at commit 807da64 recorded no executed project steps, so this review does not establish installation or renewal success.

Full review →
134

EasySpider

NaiboWang/EasySpider · ★ 45k ▲16/7d · JavaScript

EasySpider is a Chinese-first visual web scraper with a bilingual README, English source instructions, and an English tutorial playlist. You select information and browser actions on a page, save the workflow, and run it again without writing the whole scraper yourself.

SetupDocsCommunityMaturity

Verdict Our EasySpider run installed in 45 seconds but reported 45 dependency vulnerabilities, including 1 critical and 39 high, and completed neither a build nor a test step. Try it for supervised visual collection where someone checks the resulting data.

Full review →
135

maigret

soxoj/maigret · ★ 40k ▲177/7d · Python

Maigret is a username-search and reporting tool for OSINT investigations. Its README lists 5,900 supported sites, with the 500 highest-ranked sites checked by default. It can extract profile details and follow discovered identifiers into further searches. Basic searches require no API key. Treat matches as leads: a shared username alone does not establish that accounts belong to the same person.

SetupDocsCommunityMaturity

Verdict Consider Maigret for username discovery when recursive searches and report exports fit your investigation, and verify each lead independently. Our October 2, 2026 run at b664274 recorded successful install and build steps, followed by a nonpassing test step with async-support errors.

Full review →
136

zapret-discord-youtube

Flowseal/zapret-discord-youtube · ★ 35k ▲251/7d · Batchfile

Zapret Discord YouTube is a Russian-language Windows bundle for working around provider-level blocking of Discord, YouTube, and other configured services. The main documentation is in Russian and no English guide is included; it packages Batch scripts, traffic-routing strategies, WinDivert, and binaries from the upstream zapret projects.

SetupDocsCommunityMaturity

Verdict Our 2026-08-26 Debian lab with 3 CPUs and 8 GB of RAM could not run this Windows Batch and WinDivert bundle, so no independent install or test result exists. It is a practical option for Russian-speaking Windows users who can inspect binaries, accept administrator-level network changes, and test provider-specific strategies.

Full review →
137

CloakBrowser

CloakHQ/CloakBrowser · ★ 33k ▲150/7d · Python

CloakBrowser wraps a custom Chromium binary for Python and JavaScript automation through Playwright or Puppeteer. The project says it modifies browser fingerprints at the C++ source level. The wrapper source and distributed browser binary have separate license terms.

SetupDocsCommunityMaturity

Verdict A successful browser launch is the first adoption checkpoint. Our October 9, 2026 run at commit 37d81bc reported 55 failed tests and 43 collection/setup errors, with browser-launch closures in the retained tail.

Full review →
138

ios_rule_script

blackmatrix7/ios_rule_script · ★ 28k ▲72/7d · JavaScript

ios_rule_script is a Chinese-first collection of traffic-routing rules, rewrite rules, and automation scripts for tools such as Quantumult X and MagicJS. No English documentation is evident in the supplied README excerpt, so English-only users should expect to translate the instructions. It solves the tedious work of gathering and organizing rules and small device-side automations from many sources.

SetupDocsCommunityMaturity

Verdict Our install completed in 14 seconds, and our build completed in 2 seconds. Use ios_rule_script if you already understand the destination clients and want a broad Chinese-language rule warehouse that was pushed today.

Full review →
139

awesome-n8n-templates

enescingoz/awesome-n8n-templates · ★ 26k ▲88/7d

Awesome n8n Templates is an English-led collection of 280+ workflow JSON files for the n8n automation platform, with translated READMEs in 13 other languages. It saves users from drawing every workflow from scratch, but it does not turn community files into maintained integrations.

SetupDocsCommunityMaturity

Verdict Our lab found no runnable ecosystem or Dockerfile; issue 201 lists 7 files rejected by n8n. Use this collection as a large reference shelf if you already know how to audit n8n nodes and credentials.

Full review →
140

phoneinfoga

sundowndev/phoneinfoga · ★ 18k ▲46/7d · Go

PhoneInfoga gathers basic phone-number information and helps search external sources for investigative leads. Consider it for supervised research if you can configure its scanners and maintain your deployment. Avoid it when you need upstream bug fixes or verified identity results.

SetupDocsCommunityMaturity

Verdict PhoneInfoga is a candidate for supervised phone-number research when your team can configure scanners and own repairs. The October 2, 2026 run at commit `55807b0` recorded installation success, but contradictory build output and package setup failures prevent us from calling the deployment validated.

Full review →
141

cli

larksuite/cli · ★ 18k ▲71/7d · Go

Lark CLI is the official command-line interface for operating Lark and Feishu services such as Messenger, Docs, Base, Calendar, Mail, and Tasks. It gives people and AI agents structured commands for work that would otherwise require the web app or custom Open Platform code.

SetupDocsCommunityMaturity

Verdict Our Lark CLI install failed after 28 seconds because its installer could not find `checksums.txt`, so commit 35bd5ec is a poor choice for unattended bootstrap despite the project's broad official coverage. It is still worth evaluating for developers and controlled agents already committed to Lark, especially when dry runs and structured output can sit behind human approval.

Full review →
142

dagger

dagger/dagger · ★ 17k ▲52/7d · Go

Trial Dagger if your team needs reusable container workflows that developers can run locally and in CI. Dagger is a software delivery automation platform with typed APIs and language SDKs. Our run on September 20, 2026, at commit 3617e2a evaluated its source checkout in our sandbox, not an application delivery pipeline. The recorded build step returned zero despite vet diagnostics; the test command returned a nonzero exit.

SetupDocsCommunityMaturity

Verdict Trial Dagger for reusable container workflows shared between developer machines and CI, then compare one representative pipeline's actual inputs and outputs in both environments. Our September 20, 2026, source run at commit 3617e2a recorded a zero build-step exit alongside vet diagnostics and a nonzero test exit.

Full review →
143

corsair

corsairdev/corsair · ★ 16k ▲107/7d · TypeScript

Corsair is a TypeScript integration layer that gives apps and AI agents one interface for calling third-party services, handling OAuth, receiving webhooks, and storing synced data. It can run inside your application, expose integrations over REST or MCP, and use Corsair Hub for hosted connection and callback pages.

SetupDocsCommunityMaturity

Verdict Our Corsair run installed 2,431 packages, timed out building after 900 seconds, and finished Jest with 19 failures, so adopt it plugin by plugin rather than trusting the whole catalog at once. Its shared TypeScript API, tenant-scoped credentials, and agent adapters solve real integration work.

Full review →
144

holehe

megadose/holehe · ★ 15k ▲75/7d · Python

Holehe is a command-line OSINT tool that checks whether an email address appears to be registered on many websites. It infers the answer from signup, login, and password-recovery responses, sometimes returning masked recovery details without intentionally sending a message to the address.

SetupDocsCommunityMaturity

Verdict Our Holehe install took 36 seconds, used 42 MB, built in 9 seconds, and had no test target, so its clean setup says little about whether 120-plus live provider probes remain correct. Use it only for authorized, low-volume reconnaissance where every hit is corroborated manually.

Full review →
145

ax

google/ax · ★ 14k ▲303/7d · Go

AX is Google's open-source control plane for running agent tasks in isolated cluster sandboxes. You describe the task, workspace, network access, and model connection in YAML, then use a Kubernetes-like CLI to apply, watch, pause, resume, or inspect the work.

SetupDocsCommunityMaturity

Verdict Our AX run installed 247 packages and passed all 16 test packages, but the deployment path still has open reports involving wildcard egress and manifest-driven command execution. Trial it in an isolated cluster if you need Kubernetes-scale agent scheduling and can review the security boundaries yourself.

Full review →
146

wx_channels_download

ltaoo/wx_channels_download · ★ 10k ▲138/7d · Go

wx_channels_download is a Chinese-first desktop utility that adds download controls to WeChat Channels by running a local proxy and modifying responses. The README and documentation are written in Chinese; we found no English guide. It can save videos, batch-download a creator's posts, handle live streams with FFmpeg, and expose local controls through an API or MCP.

SetupDocsCommunityMaturity

Verdict Our wx_channels_download run installed 459 packages, built in 139 seconds, and passed all 4 tests, giving it the cleanest lab result in this batch. Use it only if the injected WeChat workflow saves enough time to justify an administrator-run proxy and a trusted root certificate.

Full review →
147

claude-code-security-review

anthropics/claude-code-security-review · ★ 6.7k ▲111/7d · Python

Claude Code Security Reviewer is a GitHub Action that sends pull-request changes to Claude Code, filters the resulting findings, and can post security comments on affected lines. The repository also supplies the prompt behind Claude Code's built-in `/security-review` command.

SetupDocsCommunityMaturity

Verdict Our build of Claude Code Security Reviewer installed 52 packages in 17 seconds and built in 6 seconds, but the harness found no test target to run. Treat it as a second reviewer for maintainer-approved code, never as the only security gate or as a safe scanner for hostile pull requests.

Full review →
148

Telegram-Media-Downloader

Neet-Nestor/Telegram-Media-Downloader · ★ 6.4k ▲58/7d · JavaScript

Telegram Media Downloader is an English-documented browser userscript that restores download buttons in Telegram Web, including chats and channels where saving is restricted. It handles images, GIFs, videos, stories, and some voice messages without asking for a separate Telegram API account.

SetupDocsCommunityMaturity

Verdict Version 1.212 has no package-defined install, build, or test path, so reliability rests on live Telegram Web behavior and manual checking. Use it for occasional, authorized single-file saves when the browser button is the whole problem.

Full review →
149

youtube-automation-agent

darkzOGx/youtube-automation-agent · ★ 4.9k ▲257/7d · JavaScript

Consider YouTube Automation Agent, also called AgentTube, for a private trial if you want a self-hosted YouTube workflow and can review each video before publishing. The README says finished videos wait for fact review, media-rights confirmation, and your approval. Our run checked installation and the repository's test command, not a working channel production or upload.

SetupDocsCommunityMaturity

Verdict Try AgentTube privately if you need a coordinated YouTube workflow and will review every production. Our run completed installation and the test command, but did not validate live generation or publishing.

Full review →
150

phone-harness

ShawnPana/phone-harness · ★ 4.0k ▲77/7d · Python

Phone Harness connects coding agents such as Claude Code and Codex to phones through shared screen-reading and input helpers. The current README describes local iPhone control through Mac iPhone Mirroring, local Android control over adb from macOS, Linux or Windows, and hosted Android and iPhone options. Those documented capabilities extend beyond what our recorded sandbox run verified.

SetupDocsCommunityMaturity

Verdict Phone Harness is worth evaluating for supervised agent control of a phone when its documented transport fits your setup. Our October 1, 2026 run at 88b3642 completed install and build steps and produced passing pytest output.

Full review →
151

WeChatBridge

freestylefly/WeChatBridge · ★ 1.9k ▲242/7d · Swift

WeChatBridge sends chat archives you select in WeChat to AI agents, Obsidian, or a local folder through the macOS share menu. Its primary README is Chinese, with a separate English README and bilingual interface. It preserves the exported files without reading or decrypting the WeChat database.

SetupDocsCommunityMaturity

Verdict Our lab did not run WeChatBridge because Swift was unsupported by the harness and the repository had no Dockerfile. Its documented export-to-note workflow is worth a Mac trial when you need selected conversations rather than a complete chat mirror.

Full review →
152

alchemy

alchemy-run/alchemy · ★ 1.8k ▲66/7d · TypeScript

Alchemy is an infrastructure-as-code framework that lets TypeScript developers define cloud resources and the application code using them in one Effect program. It handles deployment state, resource bindings, permissions, local development, and CI for AWS and Cloudflare without a separate YAML configuration layer.

SetupDocsCommunityMaturity

Verdict Our Alchemy checkout installed 2,494 packages and occupied 3,246 MB, then both build and test stopped because Bun 1.3.13 was absent, so a Node-only team should not adopt it casually. Effect users building on AWS or Cloudflare should trial it on a disposable stack because its typed bindings can remove real glue code.

Full review →
153

FreePEP

siknet/FreePEP · ★ 1.4k ▲80/7d · Python

FreePEP is a Chinese-first Python downloader that turns page images from the People's Education Press textbook site into local PDFs. A current English README exists, and both guides explain the Web interface, command-line filters, batch mode, Chromium requirement, and restrictions on use.

SetupDocsCommunityMaturity

Verdict Our FreePEP run installed 53 packages and built in 40 seconds combined, but it had no test target, so the clean pip-audit result is stronger evidence than the skipped quality check. Try it for a small personal download if the README's copyright and rate limits fit your use, then confirm that the PDF exists and opens.

Full review →
154

autoshorts

JayWebtech/autoshorts · ★ 1.4k ▲95/7d · Rust

AutoShorts is a desktop app that finds candidate moments in long recordings and cuts them into vertical clips. It keeps project data in local SQLite, while transcription and moment ranking can run through local tools or outside AI services.

SetupDocsCommunityMaturity

Verdict Our AutoShorts checkout installed 80 npm packages in 24 seconds and built in 19 seconds, but it supplied no test target for the workflow that chooses and renders clips. Try it if local project storage and automated candidate ranking would save you a rough viewing pass, and you can inspect every cut before publishing.

Full review →
155

golive-skill

mikehasa/golive-skill · ★ 1.3k ▲25/7d · TypeScript

GoLive is a coding-agent skill and command-line tool that plans and applies app launches through your own hosting, database, DNS, and service accounts. It records created resources and verification results so you can inspect unfinished work, check for drift, and plan cleanup.

SetupDocsCommunityMaturity

Verdict Our GoLive source install failed after 9 seconds because pnpm refused esbuild@0.28.2's build script, leaving the build and deployment workflow unverified by our lab. Try its documented prebuilt skill only in disposable provider projects until you have inspected the approval and cleanup behavior yourself.

Full review →
156

jev-browser-use

wy-coliney/jev-browser-use · ★ 1.2k ▲8/7d · JavaScript

Jev Browser Use is a Codex skill and plugin that delegates routine browser clicks, navigation, toggles, and scrolling to TypeSafe's Jev model. Codex keeps responsibility for planning, typing, visual judgment, sensitive actions, and the final verification while the helper works through an existing Computer Use browser connection.

SetupDocsCommunityMaturity

Verdict Our lab produced no build or test result for Jev Browser Use because its JavaScript layout was unsupported and no Dockerfile supplied a runnable path. Try it only if Codex Computer Use already works in your task and you are comfortable sending bounded accessibility state to a Jev provider.

Full review →
157

MatrixMedia

hanliang97/MatrixMedia · ★ 1.2k · JavaScript

MatrixMedia (矩媒) is an Electron application and command-line tool for distributing videos across Chinese social platforms. The supplied README and MCP guide are in Chinese. It combines desktop publishing with commands that reuse local account sessions.

SetupDocsCommunityMaturity

Verdict MatrixMedia warrants a supervised evaluation for its documented Chinese platforms if you can maintain account sessions and verify submissions in destination dashboards. The recorded installation pulled 2,301 dependencies, but source packaging failed after 65 seconds, leaving desktop operation and publishing unverified.

Full review →
158

company-brain

supermemoryai/company-brain · ★ 1.1k ▲14/7d · TypeScript

Company Brain is a Slack assistant that remembers team conversations, answers questions using that history and connected tools, and can carry out work such as opening issues. You deploy its code to your own Cloudflare account, while its memory lives in Supermemory and its answers use a model provider you choose.

SetupDocsCommunityMaturity

Verdict Our Company Brain build passed in 8 seconds, but the test command exited 1 because two files failed to load despite 34 passing tests. Trial it with a small Slack team if Cloudflare and hosted Supermemory fit your infrastructure.

Full review →
159

macos-harness

browser-use/macos-harness · ★ 973 ▲8/7d · Python

macOS Harness gives a coding agent direct Python access to screenshots, keyboard and pointer events, Accessibility data, Apple Events, browser control, files, and shell commands on a Mac. Instead of shipping a tool for every application, it exposes a small set of low-level primitives and lets the agent write missing automation during the task.

SetupDocsCommunityMaturity

Verdict Our Debian run built macOS Harness in 7 seconds, then passed 10 of 26 tests because the remaining 16 required unavailable macOS ApplicationServices bindings. Try it on a dedicated Mac when an agent needs to bridge browser and native applications, and treat every permission as production access.

Full review →
160

backlink_skills

flaqai/backlink_skills · ★ 844 ▲41/7d · Python

Backlink Skills is primarily documented in Simplified Chinese, with a full English README and shorter documentation in many other languages. It is a collection of Codex workflows for qualifying product directories, recording authorized submissions, and writing SEO content for websites, LinkedIn, Medium, and WeChat.

SetupDocsCommunityMaturity

Verdict Our lab could not run backlink_skills because it found no supported executable ecosystem and no Dockerfile, so this review cannot offer an independent install or test result. The quality-first V2 workflow is worth borrowing for a small, authorized directory campaign because it treats relevance, approval, evidence, and ambiguous outcomes seriously.

Full review →
161

gongwen-gbt9704-skill

mizzlelover/gongwen-gbt9704-skill · ★ 834 ▲11/7d · JavaScript

Gongwen GB/T 9704-2012 Skill is a Chinese-language document skill that turns Chinese formal material into editable DOCX files following China's party and government document layout standard. The skill metadata has a short English description, but the README, operating guide, checklists, and audit records are written in Chinese; there is no full English guide.

SetupDocsCommunityMaturity

Verdict Our lab could not run commit 6a5e47a because it exposed no supported ecosystem and no Dockerfile, so we cannot independently confirm its generator or verification claims. The repository is still worth a controlled trial for Chinese-speaking teams that need editable GB/T 9704 DOCX files and understand the manual review boundary.

Full review →
162

WA-AKG

mrifqidaffaaditya/WA-AKG · ★ 813 ▲40/7d · TypeScript

WA-AKG turns one or more WhatsApp accounts into a self-hosted dashboard and REST API by connecting through Baileys. It handles sessions, messages, recurring schedules, broadcasts, auto-replies, media, and webhooks, with an optional n8n community node for workflow automation.

SetupDocsCommunityMaturity

Verdict Our WA-AKG run built successfully in 87 seconds, but its 1,343-package install had 58 known vulnerabilities and the repository supplied no test target. It is a credible starting point for a small operator who accepts Baileys, owns the database and session state, and can test every messaging path before use.

Full review →
163

bettergi-scripts-list

babalae/bettergi-scripts-list · ★ 690 ▲54/7d · JavaScript

bettergi-scripts-list is a Chinese-language community catalog of automation scripts for BetterGI and Genshin Impact. The repository has no English documentation; it collects combat routines, map paths, JavaScript tasks, and card-game strategies that run through the separate BetterGI Windows application.

SetupDocsCommunityMaturity

Verdict Our ArtifactScore run installed 10 packages in 18 seconds, but the repository supplied no build or test target, so that clean audit does not establish that any game automation works. Use this catalog only if you already accept BetterGI's Windows, Chinese-interface, display, and account-risk constraints.

Full review →
164

ARES

Mafifrizi/ARES · ★ 651 ▲11/7d · Python

ARES is a self-hosted platform for authorized red-team and purple-team engagements. It combines scoped offensive modules, campaign controls, attack-path graphs, credentials and evidence storage, operator roles, reports, an API, and an optional MCP gateway in one system.

SetupDocsCommunityMaturity

Verdict Our ARES run installed 196 packages, built in 9 seconds, and reached only 22% of its tests before the 900-second limit, while pip-audit found 1 known vulnerability. That is enough to justify a lab evaluation, not a client engagement.

Full review →
165

genspark-claw

breakstageaxe61/genspark-claw · ★ 538 · JavaScript

Genspark-Claw is a community skill pack and local CLI for adding research, reports, slide outlines, browser guidance, and call preparation to existing agents. Its MCP server lets clients retrieve and validate those instructions, while the host agent supplies the actual tools that carry them out.

SetupDocsCommunityMaturity

Verdict Our Genspark-Claw run passed all 12 tests in 3 seconds after an 8-second install that added 0 packages and used 1 MB. It is worth trying as an editable instruction pack for an agent you already operate.

Full review →
166

mcp-server

sv-number/mcp-server · ★ 533 · JavaScript

SV Number MCP Server lets an AI agent rent a temporary phone number, wait for its SMS verification code, and close or cancel the order. It exposes nine tools to MCP clients such as Claude Code, Codex, Cursor, and Claude Desktop, while the actual number inventory comes from a paid hosted service.

SetupDocsCommunityMaturity

Verdict Our SV Number MCP Server run installed 101 packages, built in 13 seconds, and reported 0 known vulnerabilities, but there was no test target to exercise its nine tools. Use it for legitimate, supervised temporary SMS verification when broad country inventory matters.

Full review →
167

omniget

OpenSelena/omniget · ★ 448 ▲55/7d · Rust

OmniGet is a desktop download manager for online video, courses, galleries, torrents, and direct files, with readers and conversion tools built into the same app. It puts yt-dlp, FFmpeg, gallery-dl, and several native extractors behind a graphical interface for people who do not want to manage command flags.

SetupDocsCommunityMaturity

Verdict Our OmniGet run installed 367 packages and cleared the build and tests in 93 seconds total, but npm audit found 9 known vulnerabilities. Use it when a graphical home for yt-dlp, course files, and local media is worth testing against your usual sites.

Full review →
168

huashu-mac-use

alchaincyf/huashu-mac-use · ★ 330 ▲3/7d · Swift

huashu-mac-use is primarily a Chinese-language Agent Skill for controlling macOS applications that lack a normal API, with an English summary at the end of its README. It gives Claude Code, Codex, and other skill-aware agents a Swift command-line kernel for reading windows, choosing an automation channel, writing with safety checks, and saving screenshots as evidence.

SetupDocsCommunityMaturity

Verdict Our lab did not run commit 4dda98c because it requires a supported Swift-on-macOS environment and provides no Dockerfile, so its impressive safety design remains unverified by us. Try it on a non-sensitive Mac if you value background reads, explicit refusal states, and screenshot evidence more than pure visual clicking.

Full review →
169

arc-cua

shhivv/arc-cua · ★ 324 · Python

arc-cua bundles two related tools: a macOS driver that lets an agent inspect and control one app window, and a decision-model loop that carries out bounded desktop tasks. The driver works through Python or MCP without a model, while the action loop can use JEV through TypeSafe or a compatible custom provider.

SetupDocsCommunityMaturity

Verdict Our arc-cua run built in 2 seconds and passed 360 tests, but 8 macOS-dependent tests failed on Debian and pip-audit reported 2 known vulnerabilities. The standalone driver is worth a guarded macOS trial when background window control and MCP access solve a specific problem.

Full review →
170

PowerShell

PowerShell/PowerShell · ★ 56k · C#

PowerShell is a command-line shell, scripting language, and framework for processing cmdlets. Its README describes cross-platform automation for Windows, Linux, and macOS, with a focus on structured data such as JSON, CSV, and XML, REST APIs, and object models.

SetupDocsCommunityMaturity

Verdict PowerShell is worth evaluating for PowerShell 7+ automation involving structured data and multiple operating systems. That recommendation rests on the documented role of the product, not a successful lab installation.

Full review →
171

sim

simstudioai/sim · ★ 30k ▲29/7d · TypeScript

Sim is a workspace for building, deploying, and managing AI agents and workflows. Its README describes visual, conversational, and code-based building, with shared files, tables, knowledge, schedules, and run logs. Our recorded source-checkout run did not verify those application features or the published-image deployment.

SetupDocsCommunityMaturity

Verdict Sim is worth evaluating when a shared agent-workflow workspace matches your team's needs and someone can own its deployment. Our October 1, 2026 source run at 7cf57f3 completed dependency installation, then encountered a desktop build failure and two script-test failures.

Full review →
172

generator-jhipster

jhipster/generator-jhipster · ★ 22k · TypeScript

JHipster is a development platform for generating web applications and microservice architectures. Its website lists Angular, React and Vue clients and Spring Boot among its backend options. This review focuses on generator-jhipster and whether its generated application conventions suit a Java team's needs.

SetupDocsCommunityMaturity

Verdict Shortlist JHipster when its application conventions match the stack your team wants and you can maintain the generated code. In our October 1, 2026 run at c6bf8ef, repository dependency installation and the separate build succeeded, but the test command exited 255 after 204 seconds.

Full review →
173

ego-lite

citrolabs/ego-lite · ★ 21k ▲1.4k/7d · JavaScript

ego lite is a Chromium browser for people and coding agents to work in parallel Spaces. The current README lists macOS and Windows support, with Linux on the roadmap. This repository contains the MIT-licensed ego-browser helper and agent skill; the browser application is a separate free download. Spaces separate work, but should not be assumed to isolate account credentials.

SetupDocsCommunityMaturity

Verdict Consider ego lite for supervised browser tasks when parallel Spaces and access through your existing coding agent fit the workflow. The current README lists macOS and Windows support.

Full review →
174

py12306

pjialin/py12306 · ★ 15k · Python

py12306 is a Chinese-language Python assistant for buying train tickets through China's 12306 service; the repository has no English README. It watches dates, routes, trains, and seat types, then can place an order for configured passengers and send notifications, with optional Redis clustering and a web page.

SetupDocsCommunityMaturity

Verdict Our py12306 run installed 35 packages in 16 seconds, built in 1 second, and found no test target, so the environment is easy to reproduce while the booking path remains unverified. Use it as a Chinese-language codebase to study or repair, not as a dependable unattended buyer today.

Full review →
175

firstmate

kunchenguid/firstmate · ★ 8.4k ▲130/7d · Shell

firstmate is a directory of instructions and shell tools that turns one terminal coding agent into a supervisor for several coding agents. It gives each worker a separate Git worktree, watches their progress, and brings one human the resulting pull requests, local merges, or research reports.

SetupDocsCommunityMaturity

Verdict Our Firstmate run installed 35 packages in 16 seconds and built in 4 seconds, but pytest collected 0 tests and exited 5, so that clean setup result does not validate the crew workflow. Use it if you already trust terminal coding agents and want one of them to supervise isolated workers under explicit merge rules.

Full review →
176

Sequoia-X

sngyai/Sequoia-X · ★ 8.4k ▲91/7d · Python

Consider Sequoia-X if you want editable Python screening rules for mainland China's A-share market and can operate the data updates and Feishu notifications yourself. Sequoia-X screens A-shares and sends picks to Feishu. Our sandbox results support trying the code, but do not establish that a scheduled market-data update or real notification delivery works.

SetupDocsCommunityMaturity

Verdict Sequoia-X is worth a controlled trial for Python users who want an A-share screener with Feishu output. Our October 8 run at commit 444c0db recorded successful command outcomes and a stored parser summary of 9 passed and 0 failed.

Full review →
177

xuexitongScript

chaolucky18/xuexitongScript · ★ 3.5k ▲192/7d · JavaScript

xuexitongScript is a browser script for playing videos and moving between sections on Chaoxing's Xuexitong course pages, with documentation primarily in Chinese and no English guide found in the repository. It reduces repeated playback and navigation clicks by acting on the course page inside your existing browser session.

SetupDocsCommunityMaturity

Verdict xuexitongScript did not run in our sandbox because our harness lacked a supported ecosystem and the repository had no Dockerfile, so its live playback reliability remains unmeasured by us. Consider it for supervised, permitted page automation if you can read the Chinese guidance and inspect the source.

Full review →
178

kueue

kubernetes-sigs/kueue · ★ 3.1k ▲17/7d · Go

Kueue is a set of Kubernetes APIs and a controller for job queueing. It controls when jobs may create pods and when active pods should stop. Its README lists priorities, fair sharing, cohorts, preemption and integrations with existing job APIs.

SetupDocsCommunityMaturity

Verdict Kueue is worth evaluating when shared Kubernetes workloads need explicit admission policies. Its documented capabilities fit that question, but our October 2, 2026 run at commit 5cb82c5 did not validate cluster operation: the build-labelled step exited successfully, while the test command failed and its causes remain unresolved.

Full review →
179

jev-trader

jarrodwatts/jev-trader · ★ 3.1k ▲23/7d · TypeScript

Jev Trader is a Bun service that reads the Kuru MON-USDC order book on Monad and chooses a buy or sell quote for each block. It can use TypeSafe's hosted Jev model, while the default mock mode uses a momentum heuristic and simulates fills without a private key.

SetupDocsCommunityMaturity

Verdict Our Jev Trader run installed 139 packages and used 120 MB in 15 seconds, but the repository supplied no build or test target, so live-money readiness is unproved. Use it to inspect a compact Monad and Kuru execution loop in dry-run mode.

Full review →
180

lanshu-create-ai-presenter-video

cclank/lanshu-create-ai-presenter-video · ★ 2.8k ▲35/7d · Python

The primary README is in Chinese, and English operational documentation is available in `SKILL.md` and its reference files. Lanshu Create AI Presenter Video is a Codex skill that organizes a topic or script, an authorized adult presenter image, voice generation, presenter animation, lip sync, captions, editing, rendering, and quality checks into one resumable job.

SetupDocsCommunityMaturity

Verdict Our install added 35 packages in 11 seconds and the build passed in 4 seconds, but no test target exercised an actual video job. Lanshu is useful as a cautious production checklist for Codex users who already own the missing generation services and can review paid outputs.

Full review →
181

agent-teams-ai

777genius/agent-teams-ai · ★ 2.4k ▲47/7d · TypeScript

Agent Teams AI is a desktop workspace for coordinating coding agents across supported runtimes. Its README describes a Kanban board, agent messages, task logs, an integrated terminal, code review and usage tracking. Our recorded source build completed, but we did not verify those desktop workflows or provider connections.

SetupDocsCommunityMaturity

Verdict Agent Teams AI is worth evaluating if you need a desktop view of several coding agents' tasks, messages and changes. Our 2026-10-11 run at commit 3dbca54 established successful install and build command exits, with a native rebuild error in the install log and no final test result before 900 seconds.

Full review →
182

centaur

paradigmxyz/centaur · ★ 1.4k ▲10/7d · Python

Centaur is a self-hosted platform that lets a team talk to coding agents from Slack or an API and gives each conversation a Kubernetes sandbox. It stores the conversation and execution state, brokers approved credentials, runs durable workflows, and can host Claude Code, Codex, Amp, or another command-line agent behind one shared control plane.

SetupDocsCommunityMaturity

Verdict Our Centaur checkout installed 364 pnpm packages in 20 seconds, but it offered no root build or test target to prove that the multi-service platform worked. Centaur is worth evaluating for a platform team that already runs Kubernetes and needs shared, credential-aware coding agents across chat and APIs.

Full review →
183

eve-software-factory-template

vercel-labs/eve-software-factory-template · ★ 1.2k ▲9/7d · TypeScript

The Eve Software Factory Template deploys Foreman, a four-stage coding-agent workflow that takes work from GitHub or Linear and ends with a reviewed draft pull request. It separates classification, planning, implementation, and review into different agents while reserving approval and merge decisions for people.

SetupDocsCommunityMaturity

Verdict Our Eve template install took 32 seconds, but its build failed in 8 seconds because the required `FACTORY_REPO` value was absent, and the repository supplied no test target. That is acceptable for a deployment template only if your team is already committed to Vercel Connect and can trial it against a scratch repository.

Full review →
184

rewards-farmer

User0332/rewards-farmer · ★ 1.1k ▲8/7d · Python

Rewards Farmer is a Python and Selenium project for automating Microsoft Rewards searches and dashboard tasks in Edge. Its README documents public-feed search queries, optional LLM providers, separate account profiles and Docker setup. Microsoft account risk remains part of the decision even when setup works.

SetupDocsCommunityMaturity

Verdict Rewards Farmer is a candidate for developers who accept the account warning and can supervise profile setup and task outcomes. Our October 5, 2026 run at 42ac7ec recorded partial test execution and environment errors, not a verified working Rewards session.

Full review →
185

Cybermes

Zyrexnn/Cybermes · ★ 985 ▲12/7d · Python

Consider Cybermes if you need a supervised workflow for authorized security testing, with tools available through an AI client or a standalone CLI. Our run on October 1, 2026, at commit 961a0dd recorded 10 npm audit advisories: eight high and two moderate. Installation succeeded, but build and tests were skipped. We did not test an assessment, MCP connection, scope enforcement, or deployment. Cybermes is an offensive-security assistant and automation framework for reconnaissance, vulnerability research, evidence collection, and reporting. Its documentation describes these capabilities; our sandbox run establishes installation success for the selected npm path.

SetupDocsCommunityMaturity

Verdict Evaluate Cybermes if you need its documented MCP tools or standalone orchestration and can supervise active testing. Our run established npm installation success, recorded 10 advisories, and skipped build and tests; assessment quality and scope enforcement remain untested.

Full review →
186

easyeda-agent

zhoushoujianwork/easyeda-agent · ★ 893 ▲67/7d · Go

easyeda-agent's primary README is Chinese. An English README exists, but it still labels itself v1.4.5 while the latest release is v1.5.2. The project gives AI agents typed commands, a local daemon, an EasyEDA Pro connector, and Skill or MCP interfaces for reading and editing schematics and PCBs.

SetupDocsCommunityMaturity

Verdict Our easyeda-agent npm run took 2 seconds, installed 0 packages, and had no build or test target, so it does not prove that the editor automation works. The project is worth a supervised trial for EasyEDA Pro users who value typed actions and readback, but that trial belongs on a copy of a real board with the exact desktop version you use.

Full review →
187

open-dot

composio-community/open-dot · ★ 693 ▲11/7d · TypeScript

Open Dot is a Mac app for personal AI assistants that browse websites, use connected apps, and keep working after you close the window. It gives each assistant its own browser and workspace, with approval rules for actions such as sending email or making a purchase.

SetupDocsCommunityMaturity

Verdict Our Open Dot run installed 798 packages and built successfully in 52 seconds, but supplied no runnable test target, so it earns a cautious local trial rather than unattended account access. Use it on a Mac when browser takeover and approval cards solve a specific chore.

Full review →
188

FluidUse

FluidInference/FluidUse · ★ 539 ▲13/7d · Swift

FluidUse is a Swift library for local form filling and decision models on Apple silicon. Its form workflow reads controls through macOS Accessibility, matches them to values you provide, and types into the running application. It targets assisted entry of known information rather than writing application essays.

SetupDocsCommunityMaturity

Verdict We did not run FluidUse commit 8c5b746 because our lab does not support Swift and the repository has no Dockerfile. Its documented profile-to-field workflow is worth evaluating on an Apple silicon Mac when local assisted entry is the goal.

Full review →
189

scalable-cli

ScalableCapital/scalable-cli · ★ 513 ▲1/7d · Rust

Scalable CLI is the official command-line interface for Scalable Broker accounts. It replaces browser scripting with structured commands for viewing portfolios, market data, savings activity, and supported account changes, while requiring explicit confirmation for sensitive actions.

SetupDocsCommunityMaturity

Verdict Our build failed after 127 seconds, and the test command failed after 51 seconds. Use the official v1.0.0 binaries if you are a Scalable Broker customer who values JSON output, local automation, and deliberate confirmation before account changes.

Full review →
190

turnstile-bypass

Sophomoresty/turnstile-bypass · ★ 510 ▲1/7d · Python

Turnstile Bypass is a Python helper that drives a visible Chrome window through a Cloudflare Turnstile widget or interstitial waiting room. It returns a short-lived token or keeps the cleared browser tab open, but it does not handle IP bans, rate limits, WAF blocks, hCaptcha, or reCAPTCHA.

SetupDocsCommunityMaturity

Verdict Our run installed 47 packages in 18 seconds and passed its 7-second build check, but the repository supplied no test target, so installation success says nothing about whether a live challenge will clear tomorrow. Use it only for authorized, narrow browser automation where headed Chrome is acceptable and a failed solve can stop cleanly.

Full review →
191

mobile-jev

droidrun/mobile-jev · ★ 459 ▲6/7d · JavaScript

Mobile Jev is a local operator console and CLI that lets a Jev model choose actions on a real Android phone connected through Mobilerun. It turns a plain-language goal into observed taps, text entry, scrolling, navigation, and a saved trace without using ADB.

SetupDocsCommunityMaturity

Verdict Our Mobile Jev run installed 184 packages, built in 21 seconds, and passed its tests in 7 seconds, so the local code is easier than the two-service live setup. Use it for supervised experiments on bounded Android tasks when you can verify the final screen state yourself.

Full review →
192

cloudflare-turnstile-bypass

henryzawadzki6542/cloudflare-turnstile-bypass · ★ 356 · Python

Cloudflare Turnstile Bypass is a small Python client that finds a Turnstile sitekey and asks the Peak service for a response token. It is useful only for authorized testing or automation where the site owner permits a third-party solver.

SetupDocsCommunityMaturity

Verdict Our run passed 11 of 11 tests and pip-audit found 0 known vulnerabilities, but it never contacted Peak, so it proves the wrapper rather than the promised solve. Use it only for an authorized target after reviewing what leaves your network and testing the paid service yourself.

Full review →
193

agentcore-cli

aws/agentcore-cli · ★ 322 ▲7/7d · TypeScript

AgentCore CLI is a command-line tool and interactive terminal UI for Amazon Bedrock AgentCore. The current README describes managed harnesses configured in YAML and runtimes containing application code you own. Projects declare resources and deployment targets in files, with AWS CDK handling infrastructure deployment.

SetupDocsCommunityMaturity

Verdict Evaluate AgentCore CLI when Amazon Bedrock AgentCore is already your deployment destination and your team can own the AWS permissions and infrastructure. In our October 1, 2026 run at be16c4f, dependency installation and the 0.31.0 source build completed, while the test command exited 1 and reported 20 failed tests.

Full review →
194

nft-public-mint

morsyxbt/nft-public-mint · ★ 293 · TypeScript

NFT Public Mint Sniper is a command-line tool that prepares and sends SeaDrop public-mint transactions on Ethereum, Base, and Robinhood Chain. It reads the price, fee recipient, wallet limit, and opening time from the contract so a user does not need an OpenSea login or API token.

SetupDocsCommunityMaturity

Verdict Our NFT Public Mint Sniper run installed 67 packages in 9 seconds and built in 7 seconds with 0 audit findings, but there was no test target for the code that signs real transactions. Use it only with a dedicated, minimally funded wallet after reading the roughly 150-line signing path and rehearsing on a low-stakes mint.

Full review →
195

cabbage-trading-machine

sopersone/cabbage-trading-machine · ★ 277 ▲2/7d · Python

Cabbage Trading Machine is a Python spot-trading bot that runs an RSI/EMA strategy through an existing trading framework, with backtest, paper, and live commands. It has an English README and a Russian getting-started guide, and gives developers a local strategy workflow rather than wallet-scanning or Jev inference.

SetupDocsCommunityMaturity

Verdict Our Cabbage test run timed out after 900 seconds with a failure marker visible, and pip-audit reported 1 known vulnerability, so a passing 5-second build is insufficient grounds for live trading. Consider it for paper-only study of its RSI/EMA framework integration.

Full review →
196

laya-ultrafast

ipenywis/laya-ultrafast · ★ 266 ▲2/7d · Python

Laya Ultrafast is a browser agent that fills forms and opens matching results using a local decision model on an Apple Silicon Mac. It adapts Jev Ultrafast so repeated browser choices can run locally, while a separate text model translates your goal into instructions.

SetupDocsCommunityMaturity

Verdict Our Laya Ultrafast run installed 57 packages in 13 seconds and passed all 51 tests, but the suite substitutes a fake model for local inference. Use it for supervised browser experiments on Apple Silicon with a separate check of the final page.

Full review →
197

flycoinrh

fruitflydev/flycoinrh · ★ 201 · Python

flycoinrh feeds browser screenshots into a simulation built from a measured male fruit-fly connectome, then maps descending-neuron activity back to cursor movement and clicks. It is a research-art rig with separate code that completes a Robinhood Chain token-launch form, signs a transaction, streams the roaming fly, and lets a language model narrate its telemetry.

SetupDocsCommunityMaturity

Verdict Our 609 MB environment built in 5 seconds, but pip-audit found 37 known vulnerabilities and no test target was available, so flycoinrh should stay away from a funded production wallet. Read and fork it as a candid connectome experiment: the code and README do a rare job of labeling where the fly ends and the rig begins.

Full review →
198

orc-claude-plugin

admte/orc-claude-plugin · ★ 51

The ORC8R plugin connects Claude to your ORC8R Cloud account so you can inspect infrastructure resources and request permitted changes in chat. It packages the remote connection and a short set of instructions for discovering operations and reporting what actually happened.

SetupDocsCommunityMaturity

Verdict The ORC8R plugin points Claude at ORC8R Cloud and always requests read access, while our lab did not run its authentication or operations workflow. Use it as a convenience for an ORC8R account you already operate, starting with reads.

Full review →
199

n8n-workflows

Zie619/n8n-workflows · ★ 57k ▲54/7d · Python

Zie619/n8n-workflows is an English-documented collection of workflow JSON files with a searchable web catalog. It helps users find automation examples; running a selected workflow requires a separate n8n instance and the accounts and credentials its integrations use.

SetupDocsCommunityMaturity

Verdict The catalog installed in 8 seconds and built in 1 second, but tests were skipped and pip-audit reported 57 known vulnerabilities. The pinned MongoDB Agent example also contains AI-labeled noOp nodes.

Full review →
200

terraform

hashicorp/terraform · ★ 50k · Go

Terraform is a source-available infrastructure-as-code CLI for describing, planning, and changing infrastructure. Its README describes versioned configuration, execution plans, and a resource graph that orders dependent changes. This repository contains Terraform core, including the CLI and graph engine; providers are separate plugins.

SetupDocsCommunityMaturity

Verdict Terraform merits evaluation when versioned infrastructure configuration and reviewable plans fit your team's workflow. Check its license, required providers, and state arrangements before adoption.

Full review →
201

invisible_dots

feder-cr/Jobs_Applier_AI_Agent_AIHawk · ★ 32k ▲70/7d · TypeScript

invisible_dots is a self-hosted computer agent. According to its guide, each Dot has a persistent QEMU virtual machine with a Linux desktop, files, memory, skills, and browser identities that retain cookies and logins. OpenRouter supplies the model. The README at our measured commit already describes this general computer agent, so readers arriving through the old AIHawk repository name should check their tutorial before proceeding.

SetupDocsCommunityMaturity

Verdict Consider invisible_dots for a supervised trial when persistent files and browser identities justify running a QEMU VM per agent. Our October 11, 2026 run at commit 37d58b6 built successfully, while its test command reported one failed secret-input assertion and three skipped tests.

Full review →
202

AI-Trader

HKUDS/AI-Trader · ★ 23k ▲92/7d · Python

AI-Trader is a hosted and self-hostable platform where software agents publish trading ideas, simulate trades, follow other agents, and compete in challenges. It gives each registered agent a token and paper balance, then exposes feeds, positions, notifications, and copy-trading actions through HTTP and WebSocket APIs.

SetupDocsCommunityMaturity

Verdict Our AI-Trader run installed 46 packages in 10 seconds with 0 audit findings, but the repository exposed neither a build target nor a test target for us to verify. Use it as a paper-trading and agent-coordination research project, with a human approving consequential actions.

Full review →
203

skypilot

skypilot-org/skypilot · ★ 11k ▲22/7d · Python

SkyPilot is infrastructure orchestration software for AI workloads. Its README describes one YAML task format or Python API for Kubernetes, Slurm, and 20+ clouds, with provisioning, code sync, setup commands, job execution, and log streaming. Our recorded run checked local setup steps; it did not establish that workloads run successfully across those targets.

SetupDocsCommunityMaturity

Verdict SkyPilot merits evaluation when a shared task interface across different infrastructure is the requirement. Our October 1, 2026 run at a72a881 recorded successful install and build steps, but the test command stopped during configuration import because UID 1000 could not be resolved.

Full review →
204

XianyuAutoAgent

shaxiu/XianyuAutoAgent · ★ 9.6k ▲54/7d · Python

XianyuAutoAgent is a Python automated responder for China's Xianyu marketplace. It routes buyer questions to pricing, technical or general reply agents and sends generated replies through a marketplace WebSocket connection. Its Chinese README describes conversation history and negotiation support. Our recorded setup run does not establish that live replies work reliably.

SetupDocsCommunityMaturity

Verdict Consider XianyuAutoAgent for a supervised development trial if you can maintain cookies, review prompts and accept the model-data flow. Our October 8, 2026 run at commit 540bbc2 recorded a successful dependency-install step, skipped tests and 6 pip-audit advisories across 2 packages, with unknown audit coverage.

Full review →
205

follow-builders

zarazhangrui/follow-builders · ★ 7.0k ▲54/7d · JavaScript

Follow Builders is a Claude Code and OpenClaw skill that turns a centrally maintained feed of AI podcasts, posts, and company blogs into a scheduled or on-demand digest. It solves the filtering problem for readers who want a fixed list of AI practitioners summarized inside an agent they already use.

SetupDocsCommunityMaturity

Verdict Our sandbox produced no install, build, or test result for commit 6df06d7 because the JavaScript repository matched no supported ecosystem and had no Dockerfile. Try Follow Builders if you already use Claude Code or OpenClaw and accept its fixed editorial feed.

Full review →
206

wecom-cli

WecomTeam/wecom-cli · ★ 3.3k ▲28/7d · Rust

wecom-cli is a command-line client for WeCom whose README and command guidance are primarily Chinese, with no English README linked. It lets people and AI agents work with messages, mail, documents, spreadsheets, tasks, calendars, meetings, cloud-drive files, and contacts from a terminal.

SetupDocsCommunityMaturity

Verdict Our wecom-cli installation stopped after 6 seconds with a Corepack wrapper error, so commit 59604ae never reached build or tests in our sandbox. The published CLI is still worth a guarded trial for a Chinese-speaking WeCom user who needs broad office automation, but its write paths require read-back checks and human approval.

Full review →
207

prod-FARM-IOS-Core

Git-Agni/prod-FARM-IOS-Core · ★ 1.8k ▲17/7d · TypeScript

Phone Farm iOS turns a Mac and connected iPhones into a local device-control and scheduling system. It provides live screens, remote input, recurring jobs, execution history, and a built-in TikTok workflow, while a plugin interface lets developers add other tasks.

SetupDocsCommunityMaturity

Verdict Our Phone Farm iOS install failed after 39 seconds when `node-native-ocr` hit `ERR_DLOPEN_FAILED` under Node 22.23.2, so setup needs hands-on investigation before a device ever connects. It is worth testing for a team that already owns the Apple hardware and wants the scheduler, dashboard, and TikTok workflow in one codebase.

Full review →
208

outlook-auto-register

lxf746/outlook-auto-register · ★ 1.1k ▲43/7d · Python

Outlook Auto Register is a Chinese-first Python tool for creating and managing Microsoft Outlook accounts through HTTP requests instead of a browser. A full English README exists, and the project combines batch registration, captcha handling, recovery proofs, proxy rotation, token retrieval, and account checks in one web console and CLI.

SetupDocsCommunityMaturity

Verdict Our run installed 51 packages in 24 seconds and built in 5 seconds, but commit 397165a offered no test target, so Outlook Auto Register is easy to assemble and hard to trust unattended. Use it only for authorized research where you control the proxy, captcha, and recovery infrastructure and can inspect every result.

Full review →
209

fire-your-seo-agency

leopard627/fire-your-seo-agency · ★ 986 ▲18/7d

fire-your-seo-agency is a Claude Code skill that guides an agent through technical SEO, answer-engine visibility, AI crawler access, brand consistency, and Naver search work. Its canonical operating instructions are Korean, while the README and all six reference guides have English versions for human readers.

SetupDocsCommunityMaturity

Verdict Our sandbox did not run commit eb9be9f because it has no supported ecosystem and no Dockerfile, so its audit workflow remains unverified by our lab. Use fire-your-seo-agency as a reviewable checklist for a Claude Code operator, especially on a Korean-market site, rather than as proof that an automated SEO system works.

Full review →
210

stonkfly

nftechie/stonkfly · ★ 898 ▲7/7d · Python

Stonkfly turns a rendered crypto price chart into input for a retained fruit-fly connectome simulation, then maps selected neural activity to buy, sell, or hold proposals. It can paper trade by default or submit guarded Coinbase Advanced spot orders, but its author says profitable learning has not been demonstrated.

SetupDocsCommunityMaturity

Verdict Our Stonkfly run passed 41 tests, but installing its 151 packages produced a 489 MB environment with 20 known vulnerabilities, so we would keep v0.1.0 away from live Coinbase credentials. It is a fascinating paper-mode connectome experiment with unusually clear limits and careful order guards.

Full review →
211

laya

aayushch/laya · ★ 780 ▲15/7d · Python

Laya is a local-first desktop command center that turns notifications from work services into AI-researched cards for a person to approve or dismiss. It connects email, chat, calendars, code hosts, and project trackers through a local Python engine and n8n, with local models or bring-your-own cloud keys.

SetupDocsCommunityMaturity

Verdict Our Laya install pulled 141 packages and 527 MB, but pytest stopped after 3 seconds because `pytest_asyncio` was missing and the audit found 5 known vulnerabilities. The product idea suits a technical solo user drowning in cross-platform notifications, provided the first trial uses disposable integrations and disables outbound actions.

Full review →
212

Rose

Alban1911/Rose · ★ 747 ▲26/7d · Python

Rose is a Windows skin changer for League of Legends, with English documentation. It watches champion select, downloads local skin assets, and injects the selected appearance when a match starts; the project says these changes affect local rendering and provide no gameplay advantage.

SetupDocsCommunityMaturity

Verdict Our Rose install took 10 seconds, but pytest hit an X display error and pip-audit found 51 known vulnerabilities before we ever reached a Windows League client. Try it only if local skins matter enough to justify administrator access, a self-supplied signed DLL, telemetry review, and repairs after game patches.

Full review →
213

mactap-app

jaskirat1616/mactap-app · ★ 737 ▲9/7d · Swift

MacTap is a macOS menu-bar utility that turns one, two, or three physical knocks on a compatible MacBook into shortcuts and system actions. It reads the laptop's motion sensor locally, with separate mappings for each side and optional per-app presets.

SetupDocsCommunityMaturity

Verdict We did not run MacTap because our sandbox lacks a supported Swift path and the repository has no Dockerfile, while 3 open issues include confirmed accidental triggers in release 2.1.2. Try it only on a supported MacBook, map the first gestures to harmless actions, and verify typing, trackpad, desk, and clamshell behavior before assigning Accept, shell, or system commands.

Full review →
214

Jev-cu

Sac-Y/Jev-cu · ★ 630 ▲3/7d · JavaScript

Jev-cu's main documentation is in Chinese, and there is no English README. It is a Codex Computer Use skill that sends text from a desktop accessibility tree to TypeSafe's Jev model, asks it to choose the next control and action, then lets local policy code decide whether Codex may execute the step. Screenshots stay local, which reduces what leaves the machine but rules out work that depends on visual layout.

SetupDocsCommunityMaturity

Verdict Our Jev-cu run installed 0 packages and passed all 22 tests in 3 seconds, but the useful path still depends on Codex Desktop and a TypeSafe key. Use it for a controlled macOS experiment where the target controls have clear text labels and a person can approve sensitive steps.

Full review →
215

third-hand

shhivv/third-hand · ★ 392 ▲6/7d · Swift

Third Hand is an experimental macOS assistant that turns chat instructions into clicks, typing, scrolling, and key presses inside desktop apps. It can keep separate task threads, target an app with an `@` mention, and use a background mode intended to leave your pointer and frontmost app alone.

SetupDocsCommunityMaturity

Verdict Our lab could not run Third Hand because commit 412fbb8 requires a macOS Swift environment and offers no Docker route, while the downloadable v0.1.3 sits 23 commits behind current source. Try it only on a spare or low-risk Mac workflow where you can supervise every action.

Full review →
216

wenzi-xiaohongshu-workflow

wenziai/wenzi-xhs-agent-skills · ★ 183 ▲3/7d

Wenzi's Xiaohongshu workflow is a Chinese-language collection of agent skills for planning an account, drafting posts, and reviewing results; no English README was listed in the repository we inspected. It helps a creator maintain a reusable account profile and turn publishing feedback into the next content plan. Its former GitHub address now redirects to wenzi-xiaohongshu-workflow.

SetupDocsCommunityMaturity

Verdict Our lab did not run Wenzi's skills pack because no supported ecosystem or Dockerfile was available, so neither agent behavior nor publishing outcomes were verified. The inspected instructions are useful for a Chinese-speaking creator who will keep an account profile and supply real evidence.

Full review →
217

turnstile-token

Roberto02800/turnstile-token · ★ 175 · Python

turnstile-token is an unofficial Python client and CLI that asks Clearance's hosted API for Cloudflare Turnstile tokens. It can discover sitekeys in page HTML and return browser identity data alongside a token, reducing the API plumbing your application needs to write.

SetupDocsCommunityMaturity

Verdict Our turnstile-token run passed all 49 tests, but the verify command checks token format without confirming Cloudflare acceptance. I'd use the library for a Clearance integration where the application owns identity replay and validates the final submission outcome.

Full review →
218

captcha-api

jamespolyakov9829/captcha-api · ★ 77 · Python

captcha-api is an unofficial Python client and command-line tool for Clearance's hosted Cloudflare Challenge and Turnstile solving service. It handles task submission and polling, with local cost estimates and concurrent batches, so you don't have to write that API plumbing yourself.

SetupDocsCommunityMaturity

Verdict Our captcha-api run passed all 36 tests, but its batch response drops the browser identity the README says follow-up requests need. Use the individual methods if you've chosen Clearance and can finish the transport integration yourself.

Full review →
219

repopilot

indada/repopilot · ★ 5 · TypeScript

RepoPilot is an independent, self-hosted controller built with the OpenAI Codex SDK. Its documented workflow turns selected GitHub issues and same-repository pull requests into test evidence, repair proposals and optional draft PRs. A separate Docker runner checks proposed changes; maintainers retain the merge decision.

SetupDocsCommunityMaturity

Verdict RepoPilot merits a bounded trial for maintainers who can supply trusted tests, explicit task selection and an isolated worker. Our October 5, 2026 run at commit 6a5597d completed dependency installation, build and the project test command successfully.

Full review →
220

obscura

h4ckf0r0day/obscura · ★ 30k ▲416/7d · Rust

Evaluate Obscura alongside your existing browser before switching. Our September 14, 2026 sandbox run at commit 89964b3 compiled successfully, but cargo test failed. The pinned README describes Obscura as a Rust headless browser engine for web scraping and AI-agent automation, with V8 execution and Chrome DevTools Protocol support. This run does not establish production readiness, Puppeteer or Playwright compatibility, or resource savings.

SetupDocsCommunityMaturity

Verdict Evaluate the checkout recorded on September 14, 2026, at commit 89964b3 if you can compare required workflows with an existing browser and investigate frame-related test failures. Their cause remains unknown.

Full review →
221

ars0n-framework-v2

R-s0n/ars0n-framework-v2 · ★ 864 ▲5/7d · Go

Consider Ars0n Framework v2 if you want guided bug bounty reconnaissance and can maintain a substantial Docker deployment. Our October 16, 2026, run at commit 2420507 covered the Go server; it did not validate the deployed platform. Ars0n combines a web interface, shared reconnaissance data and MCP controls around more than 50 tools. Its author describes it as support for human researchers rather than a replacement for manual hunting.

SetupDocsCommunityMaturity

Verdict Ars0n Framework v2 merits a restricted trial when you want its guided reconnaissance workspace and can maintain the deployment. Investigate the recorded vet diagnostics, supply the required test facilities and verify your chosen workflow before relying on scanner results.

Full review →
222

daily_stock_analysis

ZhuLinsen/daily_stock_analysis · ★ 67k ▲240/7d · Python

Daily Stock Analysis is a Chinese-first Python project for scheduled AI watchlist reports. Its English README describes coverage of mainland Chinese, Hong Kong, US, Japanese, Korean and Taiwan stocks, with web reports and chat or email delivery. Our recorded setup run does not establish a working installation or validate its financial analysis.

SetupDocsCommunityMaturity

Verdict Daily Stock Analysis is worth evaluating for a supervised watchlist research workflow, but our September 29, 2026 run at d3fee51 did not verify a working deployment. The install-step success flag conflicts with dependency-resolution errors, the build result has no retained command detail, and pytest stopped during setup because anyio was unavailable.

Full review →
223

wechaty

wechaty/wechaty · ★ 23k ▲53/7d · TypeScript

Wechaty is a TypeScript toolkit for writing chatbots against one shared set of contacts, rooms, messages, and login events. A separate connector, called a puppet, links that bot code to WeChat, WeCom, WhatsApp, or another supported messaging service.

SetupDocsCommunityMaturity

Verdict Our Wechaty checkout installed 1,302 packages, failed both build and test, and reported 75 known vulnerabilities, including 4 critical. Existing users with a proven puppet may still value its common bot API, but a new production project should start elsewhere unless WeChat compatibility is non-negotiable.

Full review →
224

skyvern

Skyvern-AI/skyvern · ★ 23k ▲41/7d · Python

Skyvern automates browser workflows using language models and computer vision. Its README describes a Playwright-compatible SDK, a no-code workflow builder, a local server and UI, and a managed cloud service. These are documented capabilities; our recorded run did not measure browser-task accuracy.

SetupDocsCommunityMaturity

Verdict Skyvern merits a supervised pilot if you need both AI-assisted browser actions and a workflow builder. Our October 4, 2026 run at commit 7a4f0f1 recorded successful install and build steps, followed by a test command blocked by missing aiofiles.

Full review →
225

res-downloader

putyy/res-downloader · ★ 21k ▲133/7d · Go

res-downloader, also called Aixiang Resource Downloader, is a Go + Wails desktop resource downloader built around local proxy capture. Its README lists video, audio, images, M3U8 and live streams, with Windows, macOS and Linux support. These are documented capabilities, not platform compatibility verified in our recorded run. The project has an English README, while its linked online documentation and user group are Chinese.

SetupDocsCommunityMaturity

Verdict Consider res-downloader when discovering media requests through an interactive desktop proxy is the task, and its certificate and proxy requirements fit your environment. Our October 2, 2026 run at f8ac4d9 confirmed a successful dependency step, but did not establish a successful application build or working desktop capture.

Full review →
226

semaphore

semaphoreui/semaphore · ★ 14k ▲33/7d · Go

Semaphore UI is a web interface for running Ansible, Terraform, OpenTofu, Terragrunt, PowerShell and shell tasks. Its README describes reusable task templates, schedules, inventories, variable groups, failure notifications and deployment access control. It is worth evaluating when a small infrastructure team needs shared automation through a browser.

SetupDocsCommunityMaturity

Verdict Semaphore UI is a plausible fit for a small infrastructure team that wants shared templates and schedules around existing automation. Our October 1, 2026 run at ddaec4c supports a narrower conclusion: the recorded setup and build-labelled steps completed, while go test exited 1.

Full review →
227

dagu

dagucloud/dagu · ★ 4.5k ▲64/7d · Go

Dagu is a self-hosted workflow engine for operations and internal automation. Its README describes a single binary that adds YAML-defined dependencies, schedules, retries, human tasks, logs and a Web UI around existing commands, without an external database or message broker. Our recorded run establishes command outcomes, not production readiness.

SetupDocsCommunityMaturity

Verdict Dagu is worth evaluating when the work already lives in scripts and the missing layer is scheduling, dependencies and visibility. Our September 11, 2026 run at commit 55b5c71 recorded successful install-labelled and build-labelled commands, followed by a capped Go test run with failed package results.

Full review →
228

jianying-headless

mcncarl/jianying-headless · ★ 3.9k ▲807/7d · Python

Jianying Headless is a mostly Chinese-documented automation toolkit that turns structured edit plans into editable Jianying Pro drafts, modifies copies of existing projects, and invokes the local macOS engine to export video. The Windows FFmpeg guide is in English, but the main getting-started and macOS documentation is Chinese; this is an unofficial integration that requires a tightly matched Jianying installation.

SetupDocsCommunityMaturity

Verdict Our Jianying Headless run built in 2 seconds but finished with 165 failed tests, 134 passed, 7 skipped, and 1 error, so this commit is not ready for unattended adoption. The narrow native workflow is interesting if editable Jianying handoff is nonnegotiable and you can reproduce the exact Mac environment, but every generated draft still needs real playback and export checks.

Full review →
229

fable-orchestrator

codejunkie99/fable-orchestrator · ★ 614 · Shell

Fable Orchestrator is a Codex skill that asks Claude Fable 5.1 to plan coding work, then routes implementation to GPT-5.6 Luna or DeepSeek V4 Flash agents exposed by Codex Router. It installs three skill files and leaves Codex responsible for tools, workspace changes, safety checks, and final verification.

SetupDocsCommunityMaturity

Verdict Our lab produced no install, build, or test result for commit e6345e5 because a Shell-only repository with no Dockerfile had no supported execution path. Fable Orchestrator is a compact, readable policy layer for a very specific Codex Router setup, not a self-contained agent platform.

Full review →
230

jev-gui-delegate

YUTA-fywoo/jev-gui-delegate · ★ 166 ▲42/7d · Python

Jev GUI Delegate is a Windows and Chrome automation prototype that lets Codex hand a bounded task to a local controller, which calls TypeSafe Jev for semantic choices. Its main docs are Simplified Chinese, with an English overview available. It checks results and can pause for missing input or uncertain decisions instead of asking the main model to direct every action.

SetupDocsCommunityMaturity

Verdict Our Jev GUI Delegate run installed 13 npm packages in 7 seconds, but only inside an Electron fixture, with build and tests skipped. The controller merits a controlled Windows trial if explicit task contracts and recovery matter to you.

Full review →
231

fragment-api-dev

slightbasebo/fragment-api-dev · ★ 138

Fragment API Dev is an English-language client and hosted API for buying Telegram Stars or Premium with GRAM or USDT on TON. It gives shops and bots a short Python or REST flow, but each purchase sends wallet signing material to a server operated outside your infrastructure.

SetupDocsCommunityMaturity

Verdict Our 3-CPU, 8 GB sandbox could not run commit 7500f0c because the repository has no supported project ecosystem or Dockerfile, leaving its paid purchase path unverified by our lab. The client has sensible order-state handling, but sending a wallet seed to a closed hosted service is too much trust for a general treasury wallet.

Full review →
232

supersidian

aicheye/supersidian · ★ 0 · TypeScript

Supersidian brings handwritten Supernote pages into dated Obsidian notes, adds live pen previews, and uses Claude Code for transcription and related-note links. It is a personal Linux desktop workflow whose README explicitly says it has no license and offers no support for other setups.

SetupDocsCommunityMaturity

Verdict Our Supersidian run passed all 25 tests after a 2-second install, but the README explicitly offers no license and no support for other setups. Treat it as a personal workflow to study until code-use permission and device compatibility are settled.

Full review →
233

superplane

superplanehq/superplane · ★ 8.5k ▲236/7d · Go

Consider SuperPlane for a bounded pilot if you want coding agents to take well-defined backlog tasks through verification and human review. The SuperPlane README describes a system that coordinates coding agents, source control, CI, review, approvals and feedback. Our run on October 1, 2026, at commit f9711f4 does not establish that this complete workflow works in production.

SetupDocsCommunityMaturity

Verdict Pilot SuperPlane for one clearly specified backlog task if its agent-to-review workflow matches your team's needs. Our October 1, 2026, run at f9711f4 left build errors in the retained output and a nonzero test exit; it did not validate deployment or successful task delivery.

Full review →
234

IDM_Pro_Tool

angusdevgo/IDM_Pro_Tool · ★ 2.2k ▲91/7d · C#

IDM Pro Tool is a Chinese-language Windows utility that bypasses Internet Download Manager's paid licensing by patching its executable and changing system state. Its README and interface documentation are in Chinese, with no English guide; the source performs administrator-level changes to the program binary, registry, access-control rules, and Hosts file.

SetupDocsCommunityMaturity

Verdict IDM Pro Tool v1.4.0 removes a vendor signature, patches up to 15 instruction sites, changes registry and ACL state, and can add eight vendor-host mappings, so our recommendation is do not use it. Buy Internet Download Manager if you need IDM, or choose an open-source download manager.

Full review →
235

gpt_nerver

Nervercc/gpt_nerver · ★ 1.2k ▲399/7d · Python

gpt_nerver is a Chinese-language PyQt6 desktop application for bulk ChatGPT account maintenance, registration, and Plus subscription workflows. The repository has no English guide; its Chinese documentation explains how it coordinates accounts, email sources, proxies, virtual cards, redemption codes, and several maintainer-run services.

SetupDocsCommunityMaturity

Verdict Our gpt_nerver run installed 45 packages and built successfully, but 2 of 32 tests failed in the session-refresh path that account maintenance relies on. Treat it as research material, not an account-operations product: its workflows hold sensitive credentials, depend on private services, and automate areas where provider rules can change.

Full review →
236

openwork

different-ai/openwork · ★ 24k ▲75/7d · TypeScript

OpenWork is a desktop app built on OpenCode and a shared capability layer for compatible AI agents. Its README describes local desktop use on macOS, Windows, and Linux without an OpenWork account, plus an optional MCP connection for sharing assigned skills, plugins, MCP connections, Google Workspace, and Microsoft 365 capabilities. The documented MCP exposes four tools.

SetupDocsCommunityMaturity

Verdict OpenWork merits evaluation if you want a local desktop workspace or shared capabilities across existing agents. Keep those choices separate: local desktop use does not require the hosted MCP sign-in, and Den has its own license.

Full review →
237

AutoHedge

The-Swarm-Corporation/AutoHedge · ★ 6.6k ▲55/7d · Python

Use AutoHedge to study AI agent orchestration; our evidence does not establish that it is ready to trade real money. AutoHedge's default CLI accepts a task prompt and calls the research system interactively. The README advertises autonomous Solana trading, but the supplied worker definitions configure the execution agent to generate order text without attaching trading tools.

SetupDocsCommunityMaturity

Verdict AutoHedge is worth inspecting for its agent roles and interactive research entry point. Our October 12, 2026, run at c549c79 established successful install and build steps, with tests skipped; it did not validate autonomous trading.

Full review →
238

cloudflare-turnstile-solver

biusberline/cloudflare-turnstile-solver · ★ 240 · Python

Cloudflare Turnstile Solver is a small Python client that finds a Turnstile sitekey in page markup and asks the commercial Peak API to return a response token. It does not solve Turnstile locally; its sensible use is authorized QA on pages you own or have permission to test.

SetupDocsCommunityMaturity

Verdict Our run needed 36 packages and 85 seconds to install a 288-line project, while the advertised `--url` command crashed before making a request. Do not put version 1.0.1 into CI unchanged.

Full review →
239

jev-workflow-builder

CTNicholas/jev-workflow-builder · ★ 151 · TypeScript

Jev Workflow Builder is a shared visual editor for connecting Jev decisions and language-model calls. It uses Liveblocks for collaboration and lets an application invoke a workflow through a REST endpoint. The repository presents it as a demo for developers building their own editor.

SetupDocsCommunityMaturity

Verdict Our Jev Workflow Builder build failed after 22 seconds, and npm audit reported 1 critical and 2 high vulnerabilities after a 458-package install. Use its source to study collaborative Jev workflows if Liveblocks fits your stack.

Full review →
240

undress-service

GangTailorUpgrade/CoomeRtool · ★ 1.2k ▲2/7d · Python

CoomeRtool now redirects to a repository named undress-service. Its installable Python package is a command-line downloader for Kemono and Coomer creator archives, while the README markets a self-hosted wardrobe stylist. Those are two different products, and the wardrobe server refers to code and folders that are absent from the repository.

SetupDocsCommunityMaturity

Verdict Our CoomeRtool install took 83 seconds, occupied 5,878 MB, and ended with 228 known vulnerabilities for a downloader whose own code is about 1,902 lines. Do not follow its Windows one-line installer, and do not treat the advertised wardrobe server as runnable from this tree.

Full review →
241

crypto-tax-calculator

caxete/crypto-tax-calculator · ★ 379 · Python

Crypto Tax Calculator says it can turn exchange and wallet records into capital-gains and income-tax reports. It includes import adapters, historical price lookup, transaction classification, and PDF or CSV output, but its setup path contains code we would not trust on a real machine.

SetupDocsCommunityMaturity

Verdict Our run installed 35 packages but the build exited 1 after 4 seconds, and the documented setup path executes heavily obfuscated Python before preparing a downloaded file for launch. Do not install Crypto Tax Calculator on a real machine or give it financial data.

Full review →
242

Polymarket-Telegram-Bot

Cleverfuxaqo1668/Polymarket-Telegram-Bot · ★ 170 · JavaScript

Polymarket-Telegram-Bot is the public landing repository for a hosted Telegram bot that alerts users to large Polymarket trades and offers one-tap betting and copy trading. The repository does not contain that bot: it has a README, an MIT license, and a small WebSocket feed example.

SetupDocsCommunityMaturity

Verdict Our 3-CPU, 8 GB sandbox did not run commit 8ea5e5c because the lab found no supported ecosystem and no Dockerfile, so this repository cannot substantiate the bot it advertises. Do not choose it as an open-source, self-hosted trading bot or as auditable wallet software.

Full review →
243

typesafe_register

Futureppo/typesafe_register · ★ 132 · Python

typesafe_register is a Python script that creates typesafe.ai accounts and requests API keys for them. Its README is in Chinese, and the measured checkout has no English documentation. It automates a hosted signup flow rather than running an AI model locally.

SetupDocsCommunityMaturity

Verdict Our typesafe_register run installed 35 packages in 11 seconds and built in 2 seconds, but provided no test target and no evidence of successful live registration. Treat it as a research script for this particular login flow, with an undocumented mail dependency and exposed credential output to address.

Full review →
244

Polymarket-Telegram-Bot

techcomet122583/Polymarket-Telegram-Bot · ★ 125 · JavaScript

Polymarket-Telegram-Bot is a public page for a hosted Telegram trading bot that advertises whale alerts, one-tap bets, copy trading, and a built-in Polymarket wallet. The repository does not contain that bot: it contains the README, an MIT license, and a short JavaScript WebSocket feed example.

SetupDocsCommunityMaturity

Verdict The repository contains 1 JavaScript feed example and no Telegram bot implementation, so it cannot support an independent code review of the wallet or trading claims. Use the example as a tiny WebSocket reference if it saves you time.

Full review →
245

iready-tool

Restdritower/iready-tool · ★ 110 · JavaScript

iReady Tool is a browser extension, userscript, and bookmarklet that tries to skip i-Ready lessons, alter recorded minutes, answer questions, and bypass diagnostics. Its stated use directly interferes with an educational platform's assessments and activity records.

SetupDocsCommunityMaturity

Verdict Our sandbox did not run iReady Tool because it exposes no supported runnable ecosystem or Dockerfile, and the repository has 0 releases and 0 open issues or pull requests to substantiate compatibility. Do not use it on a student account: its purpose is to falsify educational activity, and the recommended silent MSI download comes from an unrelated domain.

Full review →
246

tiktok-views-bot

pingwin010/tiktok-views-bot · ★ 61 · Python

tiktok-views-bot is a Python demo that attempts to submit playback activity through TikTok private endpoints. It targets artificial view counts, while the repository provides little guidance for turning the example into a maintained tool.

SetupDocsCommunityMaturity

Verdict Our tiktok-views-bot build failed after 3 seconds, and the public checkout omits the signer modules imported by its demo. I would skip it for any creator or production workflow.

Full review →
247

winutil

ChrisTitusTech/winutil · ★ 65k ▲284/7d · PowerShell

Chris Titus Tech's Windows Utility, usually called WinUtil, groups Windows app installation, system tweaks, configuration troubleshooting, and update controls in one interface.

SetupDocsCommunityMaturity

Verdict Our WinUtil run installed its Python title-screen tool in 13 seconds and built it in 1 second, leaving Windows system changes untested. Consider WinUtil when you can inspect its script and preset and rehearse the selected changes on a recoverable Windows machine.

Full review →
248

freqtrade

freqtrade/freqtrade · ★ 56k ▲222/7d · Python

Freqtrade is a Python crypto trading bot that can simulate a strategy and place trades on supported exchanges. It includes backtesting and strategy optimization, with Telegram or a web interface for control.

SetupDocsCommunityMaturity

Verdict Freqtrade installed successfully in our sandbox, but its test run failed at the dependency-audit check and recorded 1 known vulnerability. It is worth a dry-run evaluation for Python developers who can explain their strategy and operate an exchange integration.

Full review →
249

zeroclaw

zeroclaw-labs/zeroclaw · ★ 33k ▲32/7d · Rust

ZeroClaw is a personal AI assistant runtime that connects model providers to messaging channels and local tools. It runs on your machine and documents approval controls for actions and recurring procedures.

SetupDocsCommunityMaturity

Verdict ZeroClaw merits a constrained trial for experienced self-hosters who need messaging and local actions under an explicit permission policy. The sandbox build succeeded, but verification remained incomplete.

Full review →
250

OfficeCLI

iOfficeAI/OfficeCLI · ★ 33k ▲379/7d · C#

OfficeCLI is a command-line tool for reading, creating, and editing .docx, .xlsx, and .pptx files. Its documented JSON output and HTML or PNG previews give scripts and AI agents ways to inspect document elements and layout without Microsoft Office.

SetupDocsCommunityMaturity

Verdict Evaluate OfficeCLI for agents that need one interface across modern Word, Excel, and PowerPoint files. Our npm installation succeeded, but no build, tests, or document checks ran.

Full review →
251

conductor

conductor-oss/conductor · ★ 32k ▲17/7d · Java

Conductor is a workflow server that keeps orchestration in versioned JSON graphs while workers perform application tasks. Its README describes inspectable execution state and controls for retries, pauses and recovery. Our supplied run did not verify those server capabilities.

SetupDocsCommunityMaturity

Verdict Shortlist Conductor when inspectable JSON graphs and documented recovery controls justify operating a workflow server. Our detected Python setup succeeded, but tests were skipped and Java server behavior remains unverified.

Full review →
252

invisible_dots

feder-cr/AIHawk · ★ 32k ▲70/7d · TypeScript

The repository reached through feder-cr/AIHawk now leads to invisible_dots. Its guide describes a persistent AI agent with its own QEMU virtual machine, Linux desktop, files and browser identities that retain cookies and logins between tasks. The October 12 repository checks below do not establish that a working virtual desktop was deployed.

SetupDocsCommunityMaturity

Verdict At commit 81eb3ba on October 12, 2026, the build succeeded in 53 seconds and the test command exited 1 with one reported failed test. The evidence supports a supervised evaluation, not a conclusion that a persistent desktop works on your host.

Full review →
253

super-productivity

super-productivity/super-productivity · ★ 23k ▲144/7d · TypeScript

Super Productivity is a personal task planner that combines a daily schedule with time tracking. It imports assignments from services such as Jira and GitHub, so you can organize your own work alongside calendar events and export time sheets.

SetupDocsCommunityMaturity

Verdict Super Productivity built in 317 seconds in our sandbox, but its overall test command failed after 86 seconds despite Vitest reporting 278 passed and 0 failed. I would try the desktop client for planning imported assignments and tracking their hours.

Full review →
254

donutbrowser

zhom/donutbrowser · ★ 4.2k ▲39/7d · Rust

Donut Browser manages separate browsing profiles. Its README describes per-profile cookies, extensions, fingerprints and network settings, with local storage and optional sync. Browser automation and remote MCP come with the paid plans.

SetupDocsCommunityMaturity

Verdict The Donut Browser source build failed after 6 seconds, and its test command failed after 19 seconds. This run leaves the desktop and automation workflows unverified.

Full review →
255

BPB-Wizard

bia-pain-bache/BPB-Wizard · ★ 1.1k ▲3/7d · Go

BPB Wizard is a web and command-line helper for installing BPB Panel into a Cloudflare account. For BPB Panel v5.1.1, the panel's release instructions require the online Wizard and explicitly exclude client deployment.

SetupDocsCommunityMaturity

Verdict For BPB Panel v5.1.1, follow the documented online Wizard route; its release instructions exclude client deployment. The npm dependency installation succeeded, but deployment and CLI execution remain unassessed.

Full review →