Best Open-Source Databases & Data Tools

Databases, search engines, and data platforms — reviewed for real-world fit, not benchmarks. Ranked by our usability scores across setup ease, docs, community health, and maturity — 181 tools reviewed in full.

01

ClickHouse

ClickHouse/ClickHouse · ★ 50k ▲128/7d · C++

ClickHouse is a specialized open-source database designed for high-speed data analysis. It uses a column-oriented structure to quickly generate reports and dashboards from massive datasets in real-time, solving the problem of slow analytical queries that plague traditional row-based databases.

SetupDocsCommunityMaturity

Verdict ClickHouse is a monster of a database for real-time analytics, offering blistering speed for those who can tame it. If your business depends on slicing and dicing massive datasets instantly, and you're not afraid of a complex system, it's a top-tier choice.

Full review →

02

free-programming-books

EbookFoundation/free-programming-books · ★ 397k · Python

Free Programming Books is a community-maintained index of books, courses, interactive tutorials, playgrounds, problem sets, cheat sheets, podcasts, and screencasts that are free to access. It solves the first research problem for self-directed learners: finding legitimate material across many programming subjects and spoken languages without starting from a commercial search page.

SetupDocsCommunityMaturity

Verdict Our free-programming-books install finished in 16 seconds with 35 packages, the build passed in 5 seconds, and pip-audit found 0 known vulnerabilities, so trying its maintenance tooling is low risk. Use the project as a well-run discovery index, especially for non-English material, while bringing your own curriculum and checking each external link.

Full review →

03

scrapy

scrapy/scrapy · ★ 64k ▲94/7d · Python

Scrapy is a Python framework for crawling websites and turning responses into structured records. It coordinates requests, retries, parsing, link following, throttling, exports, and item processing so a scraper can grow beyond a loop around an HTTP client.

SetupDocsCommunityMaturity

Verdict Our Scrapy run installed 67 packages in 36 seconds and built in 11 seconds, but the suite finished with 10 failures and 71 collection or setup errors, so reproduce the async test environment before contributing. Use Scrapy for a crawler that needs queues, politeness controls, exports, and extension points across many requests.

Full review →

04

etcd

etcd-io/etcd · ★ 52k ▲31/7d · Go

etcd is a distributed key-value store for configuration, coordination, service discovery, and other small pieces of data that a distributed system must agree on. It uses the Raft consensus algorithm and exposes a gRPC API, giving applications consistent reads, transactions, watches, leases, and locks across a replicated cluster.

SetupDocsCommunityMaturity

Verdict etcd is the sensible default when a distributed system needs a small, strongly consistent source of truth and the team can operate quorum-based storage. Its release process, documentation, test infrastructure, and active issue work support serious use.

Full review →

05

open-source-ios-apps

dkhamsing/open-source-ios-apps · ★ 52k ▲106/7d

This English-language repository is a catalog of source-available apps for iOS, iPadOS, watchOS, tvOS, and visionOS. It helps Apple developers find real application code by product type, framework, platform, and age instead of learning only from isolated snippets.

SetupDocsCommunityMaturity

Verdict The catalog listed 1,672 Apple-platform projects on August 26, 2026, but our lab did not run it because it has no supported code ecosystem or Dockerfile. Use it as a broad discovery index, especially when you want to read complete app code across five Apple platforms.

Full review →

06

streamlit

streamlit/streamlit · ★ 46k ▲46/7d · Python

Streamlit turns a Python script into an interactive browser app, so a data scientist can ship a dashboard or model interface without building a separate frontend. It is best at getting an internal tool in front of users quickly, with widgets, charts, tables, and app state expressed in Python.

SetupDocsCommunityMaturity

Verdict Our Streamlit checkout installed 35 packages in 77 seconds but failed its source build after 11 seconds, so using the package is much easier than joining its development workflow. Choose it when a Python team needs a credible data app quickly and can work with full-script reruns.

Full review →

07

duckdb

duckdb/duckdb · ★ 41k ▲180/7d · C++

DuckDB is an in-process analytical SQL database that runs inside an application or from a command-line client. It lets analysts query local files such as CSV and Parquet directly, or persist data in a database file, without operating a separate database server.

SetupDocsCommunityMaturity

Verdict Our `api_spec/` install took 24 seconds and 37 MB, but pytest found 0 tests, so that clean build is evidence about the C API generator rather than proof that the DuckDB engine passed its suite. DuckDB is an excellent default for local analytical SQL over files or in-process data, provided one process owns writes.

Full review →

08

dns-blocklists

hagezi/dns-blocklists · ★ 27k ▲389/7d · Text

HaGeZi DNS Blocklists is a maintained collection of domain and IP lists for blocking ads, tracking, telemetry, scams, phishing, malware, and other unwanted network traffic. It gives DNS administrators several filtering strengths and tool-specific formats, so they can protect an entire network without configuring every device separately.

SetupDocsCommunityMaturity

Verdict Our run produced no install, build, or test result because this Text repository has no Dockerfile or supported executable ecosystem. Use HaGeZi if you already operate DNS filtering and want unusually clear choices between conservative and aggressive coverage.

Full review →

09

public-apis

public-apis/public-apis · ★ 481k ▲2.5k/7d · Python

Public APIs is a manually curated, English-language directory of free and freemium APIs across dozens of subjects. It helps developers find possible data sources and services without searching the web category by category, while showing basic facts such as authentication, HTTPS, and browser access support.

SetupDocsCommunityMaturity

Verdict Public APIs remains one of the best places to begin API discovery because it is broad, readable, and actively edited. Use it to form a shortlist, not to approve a production dependency.

Full review →

10

awesome-go

avelino/awesome-go · ★ 184k ▲656/7d · Go

Awesome Go is a community-edited directory of Go libraries, frameworks, software, and learning resources grouped by subject. It helps developers produce a shortlist without searching the whole module ecosystem, using published admission and maintenance criteria to filter submissions.

SetupDocsCommunityMaturity

Verdict Our Awesome Go build passed in 18 seconds, but 1 of 3 test packages failed after the main package spent about 360 seconds in an HTTP/2 path. The list is still one of the fastest ways to form a Go-library shortlist because its admission rules are specific and its last push was August 26, 2026.

Full review →

11

iptv

iptv-org/iptv · ★ 139k ▲576/7d · TypeScript

iptv-org/iptv is a maintained collection of links to publicly available live television streams from around the world. It solves the tedious job of finding and organizing those links by publishing ready-to-open M3U playlists, while related iptv-org projects supply channel metadata, program guides, and an API.

SetupDocsCommunityMaturity

Verdict Use iptv-org/iptv if you want a large, active public-stream directory and accept that it is an index, not a television service. The hosted playlist makes trying it unusually easy, and the passing test suite is reassuring for contributors.

Full review →

12

awesome-mac

jaywcjlove/awesome-mac · ★ 114k ▲485/7d · Swift

Awesome Mac is a curated directory of macOS software grouped by job, including writing, development, design, productivity, utilities, security, and AI. It helps Mac users discover options through short descriptions and badges for open source, freeware, App Store availability, and native apps.

SetupDocsCommunityMaturity

Verdict Our Awesome Mac build finished in 10 seconds after installing 303 packages, but there was no test target to check the generated catalog beyond that successful build. Use it as a broad discovery index, especially when multilingual listings and open-source or freeware badges save search time.

Full review →

13

elasticsearch

elastic/elasticsearch · ★ 78k ▲19/7d · Java

Elasticsearch is a distributed engine for full-text search, vector search, logs, metrics, and analytics over indexed documents. It solves the hard parts of finding relevant records quickly across data that is too large or too busy for a single database query path.

SetupDocsCommunityMaturity

Verdict Our 405.6 MB Elasticsearch checkout built its native Parquet Rust subproject in 68 seconds, but that result does not validate the Java server or a production cluster. Choose Elasticsearch when search relevance, vector retrieval, and operational analytics justify a dedicated stateful system with experienced owners.

Full review →

14

scikit-learn

scikit-learn/scikit-learn · ★ 67k ▲44/7d · Python

scikit-learn is a Python library for building conventional machine-learning workflows, from preparing data to training and evaluating models. It gives practitioners a consistent interface for common tasks without requiring them to assemble each algorithm and utility from scratch.

SetupDocsCommunityMaturity

Verdict Our build succeeded in 10 seconds, but the test run failed in 8 seconds because its editable loader could not find Ninja. Even with that source-testing rough edge, scikit-learn is an easy recommendation for conventional machine learning in Python because its scope, documentation routes, permissive license, and active maintenance are all clearly evidenced.

Full review →

15

prometheus

prometheus/prometheus · ★ 66k ▲74/7d · Go

Prometheus is a monitoring server and time-series database that regularly pulls metrics from applications and infrastructure. It stores labeled measurements, queries them with PromQL, evaluates alerting rules, and can send alerts when conditions are met.

SetupDocsCommunityMaturity

Verdict Our Prometheus run built successfully in 317 seconds, but its test step ended after 582 seconds with one failed result among 33, so operators should prefer a released binary while contributors reproduce the failing suite. Prometheus remains the sensible default for scrape-based metrics and PromQL when a team accepts that retention, high availability, and global storage are separate design work.

Full review →

16

pandas

pandas-dev/pandas · ★ 50k ▲28/7d · Python

pandas is a Python library for cleaning, reshaping, joining, and analyzing labeled or tabular data. Its DataFrame and Series types give developers a practical way to work with CSV files, spreadsheets, database results, time series, and missing values without writing every operation as a loop.

SetupDocsCommunityMaturity

Verdict Our pandas source install reached the 900-second limit without finishing, so a contributor checkout needs more setup patience than the release package suggests. Use pandas when compatibility, documentation, and a familiar DataFrame model matter more than squeezing every query for speed or memory.

Full review →

17

posthog

PostHog/posthog · ★ 40k ▲132/7d · Python

PostHog is an all-in-one product data platform for understanding what users do and changing a product based on that evidence. It combines analytics, session replay, feature flags, experiments, error tracking, logs, surveys, data pipelines, AI observability, and agent-facing tools so a team does not have to assemble each capability from a different vendor.

SetupDocsCommunityMaturity

Verdict Our PostHog install pulled 3,633 packages and occupied 3,615 MB, while the repository exposed no generic build or test target, so a self-hoster is taking on a very large system with limited evidence from our run. Product teams should still trial PostHog when analytics, replay, flags, experiments, and errors need one data model.

Full review →

18

postgrest

PostgREST/postgrest · ★ 28k · Haskell

PostgREST turns tables, views, and database functions in PostgreSQL into a REST API. It removes much of the usual controller code by making the database schema, permissions, and SQL functions define what clients can read or change.

SetupDocsCommunityMaturity

Verdict We did not run PostgREST at commit ed1b7a6 because our lab harness does not support Haskell and the repository has no Dockerfile. Use it when PostgreSQL is already the center of the system and your team is comfortable making views, functions, grants, and row policies the API contract.

Full review →

19

nats-server

nats-io/nats-server · ★ 21k ▲24/7d · Go

NATS Server is a message broker for connecting services, devices, and command-line tools through named subjects. Core NATS handles live publish-subscribe and request-reply traffic, while JetStream adds stored messages, replay, acknowledgments, key-value data, and object storage.

SetupDocsCommunityMaturity

Verdict Our NATS Server build passed in 52 seconds, but 2 of 10 test-package results failed in OCSP revocation cases after a 344-second run. NATS is an excellent shortlist choice for service messaging when teams understand that Core NATS is ephemeral and JetStream is a separate durability commitment.

Full review →

20

opentelemetry-collector

open-telemetry/opentelemetry-collector · ★ 7.6k ▲39/7d · Go

OpenTelemetry Collector receives telemetry, processes it through configured pipelines, and exports it to observability backends. It gives teams one vendor-neutral service for traces, metrics, and logs instead of deploying a separate agent for each format and destination.

SetupDocsCommunityMaturity

Verdict Our OpenTelemetry Collector run installed 55 packages, built in 27 seconds, and passed its single available test in 11 seconds. Use it when vendor-neutral traces, metrics, and logs need one programmable transit layer, and assign an owner to capacity, configuration, and upgrades.

Full review →

21

opendal

apache/opendal · ★ 5.4k ▲8/7d · Rust

Apache OpenDAL gives applications one API for object stores, filesystems, databases, cloud drives, protocols, and caches. Its Rust core turns each backend into an `Operator`, while layers add retries, timeouts, logging, metrics, and traffic limits around the same read and write calls.

SetupDocsCommunityMaturity

Verdict Our OpenDAL core run built in 102 seconds and passed all 652 tests in 57 seconds, making it a credible storage boundary for Rust systems that genuinely need several backends. Adopt it when provider choice and shared production layers repay the capability checks and service testing.

Full review →

22

markitdown

microsoft/markitdown · ★ 184k ▲2.7k/7d · Python

MarkItDown is a Python library and command-line tool that turns PDFs, Office files, images, audio, archives, and web content into Markdown for language-model and text-analysis pipelines. It keeps useful document structure where it can, but Microsoft says the output is not intended for high-fidelity conversion for human readers.

SetupDocsCommunityMaturity

Verdict Our MarkItDown run installed 76 packages and used 359 MB, then 443 of 444 tests passed before speech transcription omitted the number 5. Use it for fast, inspectable Markdown ingestion when source files are trusted and approximate structure is acceptable.

Full review →

23

redis

redis/redis · ★ 76k ▲82/7d · C

Redis is an in-memory data server for caches, sessions, queues, streams, search, documents, vectors, and several specialized data structures. It gives applications fast shared state and server-side operations without forcing every problem into plain key-value reads.

SetupDocsCommunityMaturity

Verdict Our Redis checkout run installed 40 packages and found 2 known vulnerabilities in a nested validator, but it collected 0 tests and never built the server, so it cannot support a server-readiness claim. Redis remains the broad default when an application needs several shared data structures and the team can operate stateful infrastructure.

Full review →

24

chinese-independent-developer

1c7/chinese-independent-developer · ★ 61k ▲145/7d · Python

Chinese Independent Developer is a Chinese-language directory of products made by independent developers in China. Its catalog and contribution instructions are written primarily in Chinese, and the repository has no dedicated English README. It helps readers discover working websites, apps, games, and developer products while giving makers a simple submission route.

SetupDocsCommunityMaturity

Verdict Our sandbox installed the repository in 11 seconds, built it in 5 seconds, and passed all 4 tests, so the maintenance code is unusually cheap to inspect and try. Use the catalog for discovering products from Chinese independent developers or as a dataset that you expect to clean yourself.

Full review →

25

trackerslist

ngosang/trackerslist · ★ 55k ▲61/7d

Trackerslist publishes a 20-entry best list and broader plain-text lists of public BitTorrent tracker addresses, refreshed every day. It saves torrent users and client-tool authors from collecting and sorting tracker URLs themselves; it is a data feed, not a tracker server or torrent client.

SetupDocsCommunityMaturity

Verdict Our 3-CPU, 8 GB sandbox could not run commit 31bd8e4 because the repository has no supported language ecosystem or Dockerfile, so Trackerslist is a published data feed rather than executable software. Use its 20-entry best list when you want a low-effort tracker source and can accept daily churn.

Full review →

26

hiring-without-whiteboards

poteto/hiring-without-whiteboards · ★ 52k ▲88/7d · JavaScript

Hiring Without Whiteboards is a community-maintained directory of companies that say their technical interviews resemble day-to-day engineering work. It helps job seekers find employers using work samples, pair programming, design discussions, or take-home exercises instead of trivia puzzles.

SetupDocsCommunityMaturity

Verdict Our run installed 243 packages in 11 seconds and passed the available tests in 6 seconds, so the repository is easy to check locally and useful as a starting list. Use it to discover employers, then verify each interview process before investing time in an application.

Full review →

27

metabase

metabase/metabase · ★ 49k ▲134/7d · Clojure

Metabase is a business intelligence application for exploring databases, writing SQL, building dashboards, scheduling alerts, and embedding analytics in other products. It gives people who do not write SQL a visual question builder, while analysts can use a native query editor.

SetupDocsCommunityMaturity

Verdict Our source install of Metabase commit ad801f2 failed after 67 seconds at its Bun preinstall check, so most BI teams should start with the official image, JAR, or cloud service. The product is a sensible default when non-SQL users need governed access to company data and a data team can own definitions and permissions.

Full review →

28

payload

payloadcms/payload · ★ 45k ▲123/7d · TypeScript

Payload is a TypeScript backend and headless CMS that runs inside a Next.js application. You define collections and fields in code, then get an admin interface, authentication, access control, REST and GraphQL APIs, drafts, localization, hooks, and generated types around that data.

SetupDocsCommunityMaturity

Verdict Our Payload workspace installed 2,292 packages and built in 101 seconds, but all 95 test files failed during setup before any tests ran. Payload is the best fit here for a TypeScript team that wants its CMS schema, backend logic, admin extensions, and Next.js frontend in one codebase.

Full review →

29

spark

apache/spark · ★ 44k ▲41/7d · Scala

Apache Spark is a distributed engine for processing large datasets through SQL, DataFrames, streaming jobs, machine-learning pipelines, and graph computations. It lets Scala, Java, and Python teams run one data-processing model locally or across a cluster; the README also lists an R API but marks it deprecated.

SetupDocsCommunityMaturity

Verdict Our Spark Python-path run installed 35 packages in 30 seconds and built in 13 seconds, but it found no test target and therefore ran 0 tests. Spark remains a sensible choice for teams that truly need distributed SQL, streaming, and DataFrame processing and can operate the cluster around it.

Full review →

30

polars

pola-rs/polars · ★ 40k ▲49/7d · Rust

Polars is a dataframe query engine for analyzing structured data through Python, Rust, Node.js, R, or SQL. It helps developers replace row-by-row transformations with column expressions that the engine can optimize, run in parallel, and sometimes stream without holding the entire dataset in memory.

SetupDocsCommunityMaturity

Verdict Our Polars sandbox installed 441 packages in 49 seconds, then both the build and tests failed because the linker could not find `-lpython3.11`, so source contributors need a prepared Python development toolchain. For users taking the packaged Python build, Polars is the first alternative we would test for a new analytical pipeline that can run on one machine.

Full review →

31

directus

directus/directus · ★ 38k ▲112/7d · TypeScript

Directus sits in front of a SQL database and generates REST and GraphQL APIs, a browser-based data Studio, authentication, permissions, automation flows, and an MCP server. It lets developers keep control of the schema while editors and operators work through a friendlier interface.

SetupDocsCommunityMaturity

Verdict Our Directus install added 2,195 packages and used 1,227 MB, then built in 42 seconds before the overall test command failed after 292 seconds. Directus is a strong fit when an existing SQL schema needs governed APIs and a capable data interface for non-developers.

Full review →

32

typeorm

typeorm/typeorm · ★ 37k ▲1/7d · TypeScript

TypeORM maps JavaScript or TypeScript classes to database tables and gives applications repositories, relations, migrations, transactions, and a query builder. It supports both Data Mapper and Active Record styles across Node.js and several browser, mobile, and desktop runtimes.

SetupDocsCommunityMaturity

Verdict Our TypeORM install took 29 seconds, pulled 951 packages, and occupied 533 MB, then the test command failed after 26 seconds in Mocha's module-loading path. Use it when database breadth, decorators, and the choice of Data Mapper or Active Record outweigh the cost of a large, compatibility-heavy codebase.

Full review →

33

jax

jax-ml/jax · ★ 36k ▲23/7d · Python

JAX is a Python library for writing numerical functions that can be differentiated, compiled, vectorized, and distributed across supported CPUs and accelerators. It solves the repeated work of hand-writing separate gradient, batching, and device-specific versions of the same calculation.

SetupDocsCommunityMaturity

Verdict Our JAX run installed 39 packages and built successfully, but pytest stopped after 200 collection or setup errors before any test ran. JAX is still the right choice when composable differentiation, compilation, vectorization, and sharding define the architecture, provided the team can test its exact backend and dtype paths.

Full review →

34

qdrant

qdrant/qdrant · ★ 35k ▲132/7d · Rust

Qdrant is a vector database for storing embeddings with JSON payloads, filtering them, and retrieving similar items through REST or gRPC. It supports dense, sparse, and multivector search, so one service can handle semantic retrieval, keyword-style matching, recommendation, and hybrid ranking.

SetupDocsCommunityMaturity

Verdict Our Qdrant build took 680 seconds, and `cargo test` ended with 23 failures in storage-read API cases despite 134 passes, so source adopters should reproduce that suite before shipping commit `74f3e85`. Qdrant remains a strong choice for teams that need filtered hybrid retrieval and can operate a stateful service.

Full review →

35

kafka

apache/kafka · ★ 34k ▲38/7d · Java

Apache Kafka is a distributed event-streaming system for moving and retaining records between applications. It solves the problem of many producers and consumers needing a shared, ordered log instead of a mesh of direct integrations.

SetupDocsCommunityMaturity

Verdict Our Kafka checkout installed 45 Python packages and built `committer-tools`, but pytest collected 0 tests, so our run says nothing about broker correctness. Kafka is the sensible choice when an organization truly needs a shared event log and can staff its operation.

Full review →

36

numpy

numpy/numpy · ★ 33k ▲45/7d · Python

NumPy is the core Python library for storing and working with large grids of numbers. It gives scientific and data applications a shared array format plus tools for algebra, Fourier transforms, random sampling, and connecting Python to C, C++, and Fortran code.

SetupDocsCommunityMaturity

Verdict Our build completed in 4 seconds, but our test command exited 3 with 2 setup/collection errors before running a single test. Use NumPy when Python software needs the standard foundation for numerical arrays: its scope, activity, and ecosystem position are unusually convincing.

Full review →

37

TrackersListCollection

XIU2/TrackersListCollection · ★ 32k ▲28/7d

TrackersListCollection is a bilingual English and Simplified Chinese repository of public BitTorrent tracker addresses. It publishes daily text files for general clients and comma-separated files for Aria2, saving users from collecting and cleaning tracker URLs by hand.

SetupDocsCommunityMaturity

Verdict Our lab did not run commit d169e6e because TrackersListCollection exposes no supported language ecosystem and no Dockerfile; it is a set of tracker files, not an application. Use `best.txt` when you want a low-effort public list and accept that UDP reachability is not checked by the publisher.

Full review →

38

awesome-quant

wilsonfreitas/awesome-quant · ★ 30k ▲117/7d · HTML

Awesome Quant is an English-language directory of libraries, data sources, research material, and services for quantitative finance. It does not run a strategy or price an asset; it helps readers find candidate tools across Python, R, Julia, Rust, spreadsheets, and hosted APIs.

SetupDocsCommunityMaturity

Verdict Our Awesome Quant checkout installed 51 packages in 23 seconds, built in 5 seconds, and passed all 70 tests in 7 seconds. Use it as a broad discovery index, especially when you want to look beyond Python or compare open code with hosted services.

Full review →

39

awesome-osint

jivoi/awesome-osint · ★ 30k ▲299/7d

Awesome OSINT is a curated directory of publicly available intelligence tools and research resources. It helps investigators, threat hunters, and curious beginners find starting points across search, social media, people, infrastructure, media, and geospatial research without assembling a bookmark collection from scratch.

SetupDocsCommunityMaturity

Verdict We could not run it because the repository has no supported language ecosystem and no Dockerfile. Use Awesome OSINT as a well-maintained map, not as a tested toolkit or investigation platform.

Full review →

40

chroma

chroma-core/chroma · ★ 29k ▲62/7d · Rust

Chroma is an open-source database for storing embeddings, documents, and metadata, then retrieving the closest records for a query. Python users can start it in memory or on disk, while Python, TypeScript, and Rust clients can connect to a separate local, self-hosted, or cloud server.

SetupDocsCommunityMaturity

Verdict Our Chroma install took 392 seconds, and pytest ran 0 tests because 2 collection errors stopped on a missing `hypothesis` module. Chroma remains one of the easiest ways to put retrieval inside a Python prototype, with a clean path to persistence and client-server use.

Full review →

41

mongoose

Automattic/mongoose · ★ 27k · JavaScript

Mongoose is an object-modeling layer for MongoDB applications written in Node.js, with alpha support for Deno. It adds schemas, casting, validation, middleware, indexes, model methods, and document population over the official MongoDB driver.

SetupDocsCommunityMaturity

Verdict Our Mongoose run installed 616 packages in 44 seconds, then the test suite hit the 900-second cap without finishing. Use Mongoose 9 when its schemas, middleware, and model API will save more code than they conceal.

Full review →

42

typesense

typesense/typesense · ★ 27k ▲26/7d · C++

Typesense is a self-hosted search server for typo-tolerant keyword search, filters, facets, sorting, vector search, and related product-search work. It aims to give application developers a simpler operating model than Elasticsearch and an open-source alternative to hosted Algolia.

SetupDocsCommunityMaturity

Verdict Our Typesense checkout installed only 10 API-test packages in 14 seconds, then exposed no build or test target, so this run proves little about server readiness. Typesense is worth a serious dataset trial for product and documentation search because the API is focused and the deployment choices are clear.

Full review →

43

rxdb

pubkey/rxdb · ★ 23k ▲2/7d · TypeScript

RxDB is a local-first NoSQL database for JavaScript applications, so an app can keep working with data on the device and synchronize it with a backend. Queries are reactive, which lets screens respond to database changes without a separate refresh layer.

SetupDocsCommunityMaturity

Verdict Our RxDB run installed 1,737 packages and used 1,241 MB, then all 1,447 Mocha tests passed before the overall test command failed on a missing `test:browser` script. Use it when offline writes, reactive queries, and backend choice justify that contributor weight.

Full review →

44

matplotlib

matplotlib/matplotlib · ★ 23k ▲11/7d · Python

Matplotlib is a Python library for turning data and calculations into static, animated, or interactive charts. It solves the practical problem of producing figures for scripts, notebooks, web servers, graphical applications, and publication-ready output without requiring a separate charting application.

SetupDocsCommunityMaturity

Verdict Our build succeeded in 12 seconds, but the test command stopped after 10 seconds with 5 collection errors and no tests run. Matplotlib remains an easy recommendation when exact figure control, broad output targets, and Python integration matter more than fashionable defaults.

Full review →

45

marimo

marimo-team/marimo · ★ 23k ▲92/7d · Python

marimo is a reactive Python notebook that stores notebooks as normal `.py` files and reruns dependent cells when an input changes. It tackles hidden state, awkward Git diffs, and the gap between exploratory notebooks, scripts, and small data apps.

SetupDocsCommunityMaturity

Verdict Our build passed in 31 seconds and all 82 measured Vitest tests passed, but the 1,899-package, 1,505 MB install shows that marimo is a full development environment rather than a thin notebook widget. It is the best fit for Python teams that value reproducible execution and reviewable `.py` files enough to adopt reactive semantics.

Full review →

46

vector

vectordotdev/vector · ★ 23k ▲35/7d · Rust

Vector is an observability pipeline that collects logs and metrics, transforms them, and routes them to one or more destinations. The same Rust service can run beside workloads as an agent, inside a pod as a sidecar, or centrally as an aggregator.

SetupDocsCommunityMaturity

Verdict Our Vector build was still unfinished at 900 seconds, and the test command failed after 708 seconds with no test count, so source contributors should expect a heavy Rust cycle. Use the packaged v0.58.0 release when logs and metrics need vendor-neutral routing, strong transformation, and configurable delivery guarantees.

Full review →

47

postgres

postgres/postgres · ★ 22k ▲62/7d · C

PostgreSQL is an open-source database that stores application data and lets software query or change it safely. It is designed for systems that need transactions, relationships between records, and room to define custom data types or functions.

SetupDocsCommunityMaturity

Verdict Our PostgreSQL build succeeded in 142 seconds, and its test step finished successfully in another 46 seconds. Use PostgreSQL when relational correctness, expressive SQL, and a mature server matter more than minimal operational work.

Full review →

48

datasets

huggingface/datasets · ★ 22k ▲18/7d · Python

Hugging Face Datasets is a Python library for loading, transforming, streaming, and sharing data used in machine learning. It gives local files and Hub datasets a common interface, with Apache Arrow storage and adapters for major Python data and training frameworks.

SetupDocsCommunityMaturity

Verdict Our Datasets checkout used 9,560 MB and its test suite timed out at 900 seconds after reaching 82%, so contributor work needs a roomy, patient environment even though the user API starts with one function. Use it for mixed-format ML data, Hub integration, or streaming when its caching and Arrow model fit your pipeline.

Full review →

49

vitess

vitessio/vitess · ★ 21k ▲35/7d · Go

Vitess is a distributed database layer that shards MySQL across multiple servers while giving applications one MySQL-compatible endpoint. It adds query routing, connection pooling, replication management, failover tooling, backups, and online data movement for teams that have outgrown a single MySQL deployment.

SetupDocsCommunityMaturity

Verdict Our Vitess checkout built in 252 seconds, but 7 of 43 Go test results failed after a 506-second run, so adoption should begin with failure analysis rather than a clean-build badge. Vitess is a serious choice for a database team that must keep MySQL while sharding beyond one server and can rehearse resharding, failover, and recovery.

Full review →

50

arnis

louis-e/arnis · ★ 18k ▲120/7d · Rust

Arnis converts real geography, elevation, and OpenStreetMap features into playable Minecraft Java or Bedrock worlds. You select a rectangle on a map, choose a scale and generation options, then it writes terrain, roads, buildings, and other mapped objects into a world save.

SetupDocsCommunityMaturity

Verdict Our Arnis run built in 345 seconds and all 618 tests passed, making it an unusually credible starting point for turning one selected real place into a Minecraft save. Use the release binary if you want a city, landscape, or classroom map and can accept interpretation errors in terrain and structures.

Full review →

51

sqlx

transact-rs/sqlx · ★ 18k ▲42/7d · Rust

SQLx is an asynchronous Rust toolkit for talking directly to PostgreSQL, MySQL, MariaDB, and SQLite. It keeps SQL visible while adding connection pools, row decoding, migrations, and optional compile-time checks against a real database schema.

SetupDocsCommunityMaturity

Verdict SQLx is the default recommendation for Rust teams that already know their SQL and want async access without hiding the database behind an ORM. Its compile-time query checks are genuinely useful, but they add schema and build-pipeline obligations that should be accepted up front.

Full review →

52

tantivy

quickwit-oss/tantivy · ★ 16k ▲27/7d · Rust

Tantivy is a Rust library for adding full-text search directly to an application or building a search service. It supplies indexing, query parsing, BM25 ranking, facets, aggregations, and stored fields, while leaving the network API, distributed operation, and product interface to you.

SetupDocsCommunityMaturity

Verdict Our Tantivy run built in 111 seconds and passed all 1,270 tests, the cleanest lab result in this group. Choose it when a Rust team wants search inside its own process and considers owning the schema, API, and operations an advantage.

Full review →

53

dbt

dbt-labs/dbt · ★ 14k · Rust

dbt turns SQL select statements into an ordered set of tables and views inside a data warehouse, with dependency tracking, tests, documentation, and repeatable builds around them. The repository's main branch now contains the Rust-based dbt v2 engine, while the familiar Python v1 code continues on the `1.latest` branch.

SetupDocsCommunityMaturity

Verdict Our dbt checkout installed 35 packages in 26 seconds and built in 2 seconds, but no detected test target ran, so that clean result does not justify a v2 migration. New teams on a supported warehouse should trial v2 because the single-binary direction and stricter parser solve real Python-era friction.

Full review →

54

dbt

dbt-labs/dbt-core · ★ 14k ▲40/7d · Rust

dbt Core turns SQL select statements into ordered tables and views inside a data warehouse, with dependency tracking, documentation, and data tests around them. It solves the gap between writing a useful query once and maintaining a shared transformation project. The repository is now split between the Python v1 line and a Rust-based v2 beta, so choosing the branch matters.

SetupDocsCommunityMaturity

Verdict Our dbt Core install took 23 seconds, used 36 MB, and built in 9 seconds, but the lab harness found no test target to run at commit 43aa6e8. It remains an easy recommendation for a SQL-first team that already has a warehouse and wants disciplined transformation work.

Full review →

55

domain-list-community

v2fly/domain-list-community · ★ 9.5k ▲36/7d · Go

v2fly/domain-list-community is a community-maintained source set and Go generator for turning categorized domain names into geosite routing data used by Project V software. It solves the tedious problem of collecting, labeling, deduplicating, and packaging domain rules so a compatible proxy can route traffic by service or category.

SetupDocsCommunityMaturity

Verdict Our build succeeded in 23 seconds, and all 2 tests passed. Use domain-list-community when you need the upstream, policy-neutral geosite vocabulary for Project V and are prepared to review how categories affect routing.

Full review →

56

cloudnative-pg

cloudnative-pg/cloudnative-pg · ★ 9.3k ▲29/7d · Go

CloudNativePG is a Kubernetes operator for running PostgreSQL clusters through custom resources. It handles primary and replica lifecycle, failover, rolling updates, services, backups, and related database objects while keeping Kubernetes as the source of desired state.

SetupDocsCommunityMaturity

Verdict CloudNativePG is a strong choice when Kubernetes is already your operating model and your team wants PostgreSQL expressed through the same API. Its scope, upgrade behavior, and backup transition are documented with refreshing precision.

Full review →

57

GeoLibre

opengeos/GeoLibre · ★ 7.4k · TypeScript

GeoLibre is an English-documented GIS application for opening, styling, analyzing, and sharing mapping data in a browser, desktop app, mobile app, or Jupyter notebook. Its main appeal is local browser processing, including a catalog of more than 1,000 geoprocessing tools, without sending selected files to a processing server.

SetupDocsCommunityMaturity

Verdict Our GeoLibre run passed all 8,352 tests, but installation took 550 seconds, used 2,569 MB, and left 23 npm audit findings to triage. The hosted app is an easy recommendation for local, exploratory GIS with ordinary data, while source builders and self-hosters should budget real time for dependencies and security review.

Full review →

58

tempo

grafana/tempo · ★ 5.5k ▲4/7d · Go

Grafana Tempo stores distributed traces, the records that show how one request moves through several services. It gives teams an OpenTelemetry-compatible backend built around object storage, while Grafana supplies the main interface for searching and reading those traces.

SetupDocsCommunityMaturity

Verdict Our Tempo run built in 204 seconds, but 1 of 54 tested packages failed, so the measured commit needs a test investigation before an internal build is promoted. Tempo is a sensible choice for a platform team already committed to Grafana and OpenTelemetry, especially when object storage fits its retention economics.

Full review →

59

alloy

grafana/alloy · ★ 3.5k ▲24/7d · Go

Grafana Alloy is a self-hosted collector that receives metrics, logs, traces, and profiles, processes them through configurable pipelines, then sends them to observability backends. It combines OpenTelemetry Collector components with built-in Prometheus pipelines, so teams can run one telemetry agent instead of maintaining separate collectors for each signal.

SetupDocsCommunityMaturity

Verdict Use Alloy when one team owns several telemetry signals and wants one programmable agent that speaks OpenTelemetry and Grafana's native ecosystems. Its component range and operational tooling are convincing, but that range also creates a large testing surface.

Full review →

60

google-timeline-visualizer

mahlernim/google-timeline-visualizer · ★ 3.0k ▲19/7d · Kotlin

Timeline Visualizer turns an exported Google Maps Timeline JSON file into an animated travel video. It runs as an Android app, an iPhone Safari web app, or an older Python desktop tool, keeping the Timeline file local while requesting map tiles for the places shown.

SetupDocsCommunityMaturity

Verdict Our Timeline Visualizer run installed 42 packages in 32 seconds and passed all 36 tests, making the Python path unusually easy to verify for a location-video tool. Use it when you want a local MP4 from an exported Timeline and will inspect the finished route before sharing.

Full review →

61

worldview

nasa-gibs/worldview · ★ 1.9k ▲9/7d · JavaScript

NASA Worldview is a browser-based map for exploring global satellite imagery served by NASA's Global Imagery Browse Services. It helps researchers, emergency teams, journalists, and the public compare observations over time without first downloading and processing the underlying Earth science data.

SetupDocsCommunityMaturity

Verdict Our Worldview run installed 1,680 packages, built successfully in 382 seconds, and passed all 9,477 Jest tests, which makes it a credible base for teams that specifically need NASA-style imagery browsing. Use the hosted site if its configuration and sharing tools already fit.

Full review →

62

sherlock

sherlock-project/sherlock · ★ 92k ▲483/7d · Python

Sherlock is a command-line OSINT tool that checks one or more usernames across more than 400 social networks. It helps investigators find possible public account pages, then leaves identity confirmation to the person doing the research.

SetupDocsCommunityMaturity

Verdict Our Sherlock run installed 46 packages and built in 9 seconds, but only 22 of 25 pytest cases passed cleanly. It is a fast first-pass account finder for investigators who understand that WAFs, site changes, and handle reuse create bad matches.

Full review →

63

awesome-public-datasets

awesomedata/awesome-public-datasets · ★ 79k ▲94/7d

Awesome Public Datasets is an English-language directory of public data sources, arranged by subject and published as a generated README. It helps researchers and developers find a promising source before they spend time searching individual agencies, archives, and dataset hosts.

SetupDocsCommunityMaturity

Verdict Our 2026-08-27 sandbox did not run commit f74e3e5 because it had no supported language ecosystem or Dockerfile, confirming that Awesome Public Datasets is a reading list rather than software to install. Use it to build a shortlist across unfamiliar subjects, then verify the chosen source's URL, terms, and data quality yourself.

Full review →

64

grafana

grafana/grafana · ★ 77k ▲107/7d · TypeScript

Grafana is a dashboard and observability application that queries metrics, logs, traces, and databases without requiring them all to live in one store. Teams use it to explore operational data, publish shared dashboards, and evaluate alert rules against many data sources.

SetupDocsCommunityMaturity

Verdict Our Grafana checkout consumed 2,610 MB, failed its build after 166 seconds, and never finished the test command within 900 seconds, so contributing from source requires the project's exact development workflow rather than generic monorepo commands. Use the packaged product when you need one mature interface across many existing data sources and can operate its permissions, alerts, plugins, and upgrades.

Full review →

65

pocketbase

pocketbase/pocketbase · ★ 61k ▲45/7d · Go

PocketBase is a self-hosted backend packaged as one Go executable, with an embedded SQLite database, user accounts, file storage, realtime subscriptions, an admin dashboard, and a REST-style API. It is for applications that need a useful backend on one server without assembling a database, authentication service, and admin panel separately.

SetupDocsCommunityMaturity

Verdict Our PocketBase build finished in 56 seconds, but 1 of 33 tested packages failed after a 394-second run, so the source is easy to compile while this exact checkout does not earn a clean test result. Use it when one server, one executable, and SQLite are positive constraints rather than compromises.

Full review →

66

meilisearch

meilisearch/meilisearch · ★ 59k ▲59/7d · Rust

Meilisearch is a search server that applications call over HTTP after sending it their documents. It handles full-text, typo-tolerant, filtered, faceted, geographic, semantic, and hybrid search without requiring an application team to build its own index and ranking engine.

SetupDocsCommunityMaturity

Verdict Our Meilisearch build succeeded in 551 seconds, but 887 of 993 tests failed in the following 703-second run, so this commit needs investigation before source-based deployment. The product remains a good shortlist candidate for application search when one HTTP service is preferable to a full search cluster.

Full review →

67

Summer2027-Internships

SimplifyJobs/Summer2027-Internships · ★ 47k ▲215/7d · Python

SimplifyJobs/Summer2027-Internships is a frequently updated directory of Summer 2027 technology internships, organized inside a GitHub README. It helps students find fresh openings across software engineering, product, AI and data science, quantitative finance, and hardware without checking hundreds of employer career pages individually.

SetupDocsCommunityMaturity

Verdict Our install completed in 29 seconds and the build in 8 seconds, but the real product is the 1,030-role README, not the Python package. Use it as a high-signal starting feed if you are applying for Summer 2027 tech internships, then verify every opening and eligibility rule on the employer site.

Full review →

68

milvus

milvus-io/milvus · ★ 46k ▲98/7d · Go

Milvus is an open-source database for storing and searching vectors alongside fields such as text, numbers, and JSON. It is built for search, recommendation, and retrieval systems that have outgrown an in-process index and need either a standalone server or a distributed cluster.

SetupDocsCommunityMaturity

Verdict Our Milvus build finished in 11 seconds, but 72 of 85 Go test targets failed in the same sandbox, so adopting it means accepting a serious system with a demanding contributor path. Use Milvus when vector retrieval is core infrastructure, you need its mix of dense, sparse, filtered, and full-text search, and your team already knows how to operate stateful services.

Full review →

69

data-engineering-zoomcamp

DataTalksClub/data-engineering-zoomcamp · ★ 46k ▲157/7d · Jupyter Notebook

Data Engineering Zoomcamp is a free 9-week course that teaches data engineering through recorded lessons, homework, workshops, and a final pipeline project. It takes learners through containers, cloud infrastructure, orchestration, warehousing, analytics engineering, batch processing, and streaming instead of teaching one tool in isolation.

SetupDocsCommunityMaturity

Verdict Our targeted Zoomcamp run installed 34 packages in 21 seconds, built in 4 seconds, and found no test target in `03-data-warehouse/extras/`. The course is a strong free choice for a learner who will finish a cloud pipeline and can tolerate tools changing between recorded lessons and current installs.

Full review →

70

ccxt

ccxt/ccxt · ★ 44k ▲81/7d · Python

CCXT is an MIT-licensed library that gives developers one API for market data and trading across 103 cryptocurrency exchange markets and 7 prediction-market exchanges. It covers JavaScript, TypeScript, Python, C#, PHP, Go, and Java, with an optional CLI and MCP server for terminal or agent access.

SetupDocsCommunityMaturity

Verdict Our CCXT install finished in 41 seconds with 498 packages, then both the build and test commands stopped on the same TypeScript error. Use the published package when you need broad exchange coverage and accept that exchange-specific behavior still needs testing.

Full review →

71

faiss

facebookresearch/faiss · ★ 41k ▲28/7d · C++

Faiss is a C++ library, with Python wrappers, for finding vectors that are closest to a query vector and for clustering vector data. It supplies exact and approximate indexes for workloads such as semantic search and recommendation, leaving storage, metadata, permissions, and application APIs to the developer.

SetupDocsCommunityMaturity

Verdict Our Faiss checkout installed 35 packages in 41 seconds, then its build failed and all 89 tests either failed or hit setup errors. Try the supported Conda package if you need direct control over high-scale vector indexes, but prove your chosen index and wrapper on production-like data before adoption.

Full review →

72

tidb

pingcap/tidb · ★ 41k ▲28/7d · Go

TiDB is a distributed SQL database that speaks the MySQL protocol while splitting query processing, metadata, and storage across separate services. It is built for applications that need transactions, horizontal growth, and failover beyond one database server. TiFlash adds columnar storage for analytical queries, while TiKV holds transactional rows.

SetupDocsCommunityMaturity

Verdict Our TiDB build took 385 seconds, and go test reached the 900-second cap with 4 passed and 2 failed out of 6, so this is a serious platform choice only for teams that already need distributed SQL. TiDB is a strong candidate for MySQL-oriented applications that need independent compute and storage growth plus mixed transactional and analytical work.

Full review →

73

seaweedfs

seaweedfs/seaweedfs · ★ 35k ▲181/7d · Go

SeaweedFS is a distributed storage system that can serve blobs, files, S3-compatible objects, FUSE mounts, and Iceberg tables from the same cluster. It is designed to keep metadata for huge numbers of small files away from one central bottleneck while allowing capacity to grow by adding volume servers.

SetupDocsCommunityMaturity

Verdict Our SeaweedFS build took 306 seconds, and after 900 seconds its test run showed 13 passing and 18 failing packages, with S3 tagging calls unable to reach a localhost service. SeaweedFS deserves a serious pilot when small-file scale, S3, FUSE, and flexible tiering must coexist, but it demands a storage team rather than a spare-server mindset.

Full review →

74

dragonfly

dragonflydb/dragonfly · ★ 32k ▲75/7d · C++

Dragonfly is a C++ in-memory data store that accepts Redis and Memcached clients while using multiple CPU cores on one server. It is meant for teams replacing a cache or Redis-compatible store without rewriting application calls, though its compatibility and license both have boundaries.

SetupDocsCommunityMaturity

Verdict Our Dragonfly test command exited 1 after 10 seconds with 0 tests passed and 3 collection errors, so commit 00693a0 did not clear our source-verification gate. Dragonfly still deserves a production trial for Linux teams constrained by a single Redis process, but our run did not measure its throughput claims.

Full review →

75

sequelize

sequelize/sequelize · ★ 30k · TypeScript

Sequelize is a Node.js and TypeScript ORM that maps application models and relationships to SQL databases. It gives one API for common queries, transactions, associations, eager loading, and several database engines, while still allowing raw SQL when the abstraction runs out.

SetupDocsCommunityMaturity

Verdict Our Sequelize install pulled 1,425 packages and 721 MB, then the build failed in 64 seconds because the SQLite native module required GLIBC 2.38. Use stable Sequelize 6 when its model API and wide dialect coverage fit an established Node service, but write integration tests against the exact database you deploy.

Full review →

76

awesome-datascience

academic/awesome-datascience · ★ 30k ▲42/7d

Awesome Data Science is an English-language directory of courses, books, tools, datasets, communities, and other links for learning or practicing data science. It gives newcomers a wide set of places to explore, but the repository itself is a reading list rather than a course, library, or working data-science environment.

SetupDocsCommunityMaturity

Verdict Awesome Data Science had 29,949 stars and a 2026-09-08 push, but its deliverable is a set of links rather than a runnable course. Use it as a broad index when you already know how to judge sources and are willing to verify every destination.

Full review →

77

nautilus_trader

nautechsystems/nautilus_trader · ★ 29k ▲276/7d · Rust

NautilusTrader is a Rust-based engine for researching, backtesting, and running automated trading systems across multiple markets and venues. Python can define strategies and orchestration while the event-driven core handles market data, orders, risk, execution, accounting, and simulation.

SetupDocsCommunityMaturity

Verdict Our NautilusTrader source build and test steps each hit the 900-second limit while still compiling, so this is a serious codebase that needs stronger hardware and a deliberate evaluation window. It is worth testing for a quant team that wants one event-driven architecture from research through live execution, preferably through a stable prebuilt wheel.

Full review →

78

mongo

mongodb/mongo · ★ 29k ▲14/7d · C++

MongoDB is a document database that stores application records as BSON documents and queries them through the `mongod` server. This repository contains the database server and the `mongos` router used for sharded deployments, rather than the separate shell or language drivers.

SetupDocsCommunityMaturity

Verdict Our MongoDB checkout installed 93 Node-tooling packages in 10 seconds, but exposed no build or test target, so our run says nothing about whether the 10,060,928-line C++ server compiles or passes its suites. Use packaged MongoDB when document-shaped data and its operational model suit the application; do not choose it because this repository appeared easy to install.

Full review →

79

flink

apache/flink · ★ 26k ▲13/7d · Java

Apache Flink is a distributed engine for processing live streams and bounded data sets while keeping state, event time, and failures under control. Its main APIs are Java, with PyFlink providing Table and DataStream APIs for Python teams that need the same engine.

SetupDocsCommunityMaturity

Verdict Our PyFlink run installed 35 packages in 64 seconds and built in 8 seconds, but pytest hit 87 collection or setup errors before executing one test. Flink remains a serious choice for teams that specifically need stateful stream processing and can support a JVM cluster, yet this checkout did not give Python contributors a trustworthy clean-container test path.

Full review →

80

dolt

dolthub/dolt · ★ 24k ▲44/7d · Go

Dolt is a SQL database that versions tables and schemas with Git-style commits, branches, diffs, merges, pushes, and pulls. Applications connect through a MySQL-compatible server, while developers can inspect or change database history from SQL procedures and the `dolt` command line.

SetupDocsCommunityMaturity

Verdict Dolt earns a trial when data changes need the same review and merge habits developers use for source code. It is a poor default for an ordinary CRUD application because the extra commit graph only pays off when the team will query, branch, or exchange history.

Full review →

81

dataease

dataease/dataease · ★ 24k ▲19/7d · Java

DataEase is a Chinese-first open-source business intelligence application, and English documentation is available alongside several other translations. It connects to databases, files, and APIs, then lets teams build and share charts through a drag-and-drop interface instead of writing every report from scratch. SQLBot integration adds conversational data analysis for organizations that want that option.

SetupDocsCommunityMaturity

Verdict Our install completed in 226 seconds, but the checkout gave us no build or test target to validate afterward. DataEase looks like a serious choice for Chinese-speaking teams that want self-hosted, drag-and-drop BI across many data sources, and its fresh v3.0.0 release plus same-day push are strong health signals.

Full review →

82

ddia

Vonng/ddia · ★ 24k ▲146/7d · Python

Vonng/ddia is a Chinese-language online translation of the first and second editions of Martin Kleppmann's *Designing Data-Intensive Applications*. The repository's book content and setup notes are primarily Chinese, with Simplified and Traditional Chinese editions; it points English readers to the original English book rather than providing English documentation of its own.

SetupDocsCommunityMaturity

Verdict Our DDIA site build finished in 7 seconds, but `go test ./...` found 0 packages and exited 1, confirming that this is a publishing project rather than tested Go software. Chinese-speaking readers should use the hosted edition for study and buy the official book when they need the authors' English text or a publisher-backed copy.

Full review →

83

timescaledb

timescale/timescaledb · ★ 24k ▲47/7d · C

TimescaleDB is a PostgreSQL extension for storing and querying time-series and event data without moving the application to a separate database language. It adds time-based partitioning, columnar storage, retention tools, time buckets, and incrementally refreshed aggregates while keeping SQL and the PostgreSQL ecosystem.

SetupDocsCommunityMaturity

Verdict Our TimescaleDB source setup stopped after 12 seconds because CMake could not find `pg_config`, so a bare C/C++ container is not enough to build it. Use TimescaleDB when keeping PostgreSQL is a firm requirement and time-based tables justify specialized partitioning, compression, and continuous aggregates.

Full review →

84

rocketmq

apache/rocketmq · ★ 23k · Java

Apache RocketMQ 5.5.1 is a distributed message broker for moving events between services, keeping them until consumers are ready, and supporting ordered or transactional delivery. It targets systems that need more control over routing, retries, storage, and failure handling than a process-local queue can provide.

SetupDocsCommunityMaturity

Verdict Our RocketMQ run installed in 72 seconds and built in 212 seconds, then 3 of 322 tests ended in setup or collection errors. That result supports a serious evaluation, not a casual approval: the code builds and most executed tests pass, yet the complete gate is red.

Full review →

85

openobserve

openobserve/openobserve · ★ 22k ▲244/7d · TypeScript

OpenObserve is a self-hosted system for collecting and searching logs, metrics, traces, browser monitoring data, and LLM telemetry. It stores analytical data in Parquet on object storage and puts search, dashboards, alerts, and ingest pipelines in one interface.

SetupDocsCommunityMaturity

Verdict Our OpenObserve build exceeded 900 seconds, while all 304 cargo tests passed in 195 seconds, so the code looks testable but expensive to compile on a 3-CPU box. Trial it when consolidating observability could remove several services and object storage already fits your operations.

Full review →

86

Lean

QuantConnect/Lean · ★ 22k · C#

LEAN is an open-source engine for writing, backtesting, and running algorithmic trading strategies in C# or Python across several financial markets. It handles event sequencing, portfolios, market data, brokerage connections, and local or cloud workflows so a quant developer can work inside one trading model.

SetupDocsCommunityMaturity

Verdict Our Lean run installed 35 packages in 15 seconds, occupied 37 MB, and built in 5 seconds; the harness found no test target and skipped tests. Use LEAN when the same event-driven model must cover serious backtesting and live brokerage work, and your team can operate its Docker, data, and .NET layers.

Full review →

87

machine-learning-for-trading

stefan-jansen/machine-learning-for-trading · ★ 21k ▲65/7d · Jupyter Notebook

Machine Learning for Trading is the code and notebook companion to Stefan Jansen's third-edition book on researching, testing, and operating machine-learning trading strategies. It covers the path from market data and feature design through backtesting, costs, risk, and live deployment.

SetupDocsCommunityMaturity

Verdict Our install consumed 10,031 MB, pip-audit found 12 known vulnerabilities, and the tests were still at 12 percent when our 900-second limit expired. Use Machine Learning for Trading as a serious course repository when you want its end-to-end research discipline and can isolate the environment.

Full review →

88

cube

cube-js/cube · ★ 21k ▲48/7d · Rust

Cube Core is an open-source semantic layer that defines business metrics, dimensions, joins, and access rules once, then serves them through SQL, REST, and GraphQL. It gives BI tools, embedded analytics, and AI agents the same governed meaning for a metric, but it does not include a finished dashboard interface.

SetupDocsCommunityMaturity

Verdict Our Cube Core install added 4,062 packages and 2,065 MB, then built in 16 seconds without exposing a root test target, so this is a substantial data service that needs its own correctness suite. Choose it when several consumers need the same metrics through SQL, REST, and GraphQL, and your team wants to build the surrounding analytics experience.

Full review →

89

shardingsphere

apache/shardingsphere · ★ 21k ▲2/7d · Java

Apache ShardingSphere is a database layer that gives existing databases sharding, read/write splitting, federation, security controls, traffic management, and observability without replacing them. Applications can use it as an embedded JDBC component or connect through a separately deployed database proxy, which makes it useful when a team needs distributed data features across heterogeneous systems while retaining its current databases.

SetupDocsCommunityMaturity

Verdict Our build completed in 215 seconds, and all 5,298 tests that ran passed. ShardingSphere deserves a serious evaluation when preserving existing databases is a firm requirement and the team genuinely needs sharding, federation, governance, or proxy access.

Full review →

90

mybatis-3

mybatis/mybatis-3 · ★ 20k ▲3/7d · Java

MyBatis is a Java data mapper that connects application objects to SQL statements or stored procedures written in XML or annotations. It removes much of JDBC's repetitive result handling while leaving the query and transaction choices visible to the developer.

SetupDocsCommunityMaturity

Verdict Our MyBatis checkout installed in 636 seconds, then both the build and test commands stopped because Maven 3.9.9 did not meet the 3.9.16 minimum. Use MyBatis when SQL ownership is a design requirement and your Java team will manage mapper and session discipline.

Full review →

91

VictoriaMetrics

VictoriaMetrics/VictoriaMetrics · ★ 18k ▲39/7d · Go

VictoriaMetrics is an open-source system for storing, querying, and managing time-series data such as infrastructure metrics, sensor readings, and application telemetry. It can act as long-term storage for Prometheus or as a Grafana-compatible replacement for Prometheus and Graphite, with both single-node and clustered deployment options.

SetupDocsCommunityMaturity

Verdict Our build succeeded in 105 seconds, but 2 of 190 tests failed. VictoriaMetrics deserves a serious trial if Prometheus retention, ingestion scale, or cross-instance querying has become a real operational constraint, because the open-source edition includes both single-node and cluster modes.

Full review →

92

juicefs

juicedata/juicefs · ★ 14k ▲24/7d · Go

JuiceFS is a distributed POSIX file system that keeps file data in object storage and file metadata in a separate database. Applications mount it like a shared drive, while the client translates ordinary file operations into metadata transactions and object-store reads and writes.

SetupDocsCommunityMaturity

Verdict Our JuiceFS checkout installed 550 npm packages in 36 seconds, but it exposed no build or test target and npm audit found 21 known vulnerabilities, so our run did not validate the Go file-system client. JuiceFS is worth a staged trial for shared POSIX access over object storage when a storage team can own the metadata service and recovery plan.

Full review →

93

datahub

datahub-project/datahub · ★ 13k · Python

DataHub is a catalog that collects facts about an organization's tables, dashboards, pipelines, models, and owners. It gives people one place to search data, trace where it came from, apply governance rules, and react when metadata changes.

SetupDocsCommunityMaturity

Verdict Our DataHub Actions run passed 191 of 204 tests, while 13 PgQueue cases failed, so DataHub is best for a staffed data platform team that can validate each part before rollout. The catalog covers a wide operational problem and the repository is very active, but the 8 GB multi-service quick start and staged v1.7 upgrade path make self-hosting a platform commitment.

Full review →

94

cocoindex

cocoindex-io/cocoindex · ★ 12k ▲41/7d · Rust

CocoIndex is a Python framework with a Rust core for keeping derived data in sync with changing sources. You declare how files, databases, messages, or media become rows, vectors, graph nodes, or files, and it tracks what must be recomputed when either the input or transformation changes.

SetupDocsCommunityMaturity

Verdict Our CocoIndex build passed in 9 seconds, but its test run finished with 309 passed, 192 failed, 125 skipped, and 8 setup or collection errors, so adoption should begin with the exact connectors you will operate. It is a strong candidate when re-embedding or re-extracting unchanged data is a real cost and Python is the primary authoring language.

Full review →

95

datafusion

apache/datafusion · ★ 9.3k ▲17/7d · Rust

Apache DataFusion is a Rust query engine for developers building databases, analytics systems, and data pipelines. It supplies SQL and DataFrame interfaces over Apache Arrow data, with extension points for planners, file formats, functions, and execution operators.

SetupDocsCommunityMaturity

Verdict Our DataFusion harness installed 35 Python packages in 16 seconds, but its build failed 4 seconds later and no tests ran, so that run does not establish a working Rust build. DataFusion remains a strong candidate for teams that specifically need an embeddable Arrow-based query engine and can evaluate it through the documented Cargo path.

Full review →

96

wealthfolio

wealthfolio/wealthfolio · ★ 9.0k ▲139/7d · Rust

Wealthfolio is a local-first portfolio and personal-finance tracker for investments, net worth, spending, goals, and scenario planning. The free app stores data in a local SQLite database and supports manual entries or CSV imports; an optional paid service adds read-only brokerage connections and encrypted multi-device sync.

SetupDocsCommunityMaturity

Verdict Our Wealthfolio run passed all 1,504 tests in 261 seconds, but the 37-second frontend build failed on missing UI declarations and a TypeScript type error. The downloadable app is a strong candidate for investors who value local storage and will reconcile imported balances against broker statements.

Full review →

97

dbmate

amacneil/dbmate · ★ 7.4k ▲36/7d · Go

Dbmate is a standalone database migration command that lets teams keep schema changes in timestamped plain SQL files. It gives services written in different languages one shared way to create databases, apply or roll back migrations, track status, and save a schema dump.

SetupDocsCommunityMaturity

Verdict Our Dbmate build succeeded in 83 seconds, but 5 of 17 test groups failed and the log tail identified SQLite dump and FTS5 cases, so SQLite teams should reproduce the suite before adopting it. Dbmate remains a good fit for polyglot teams that want plain SQL, one small CLI, and Git-visible schema dumps.

Full review →

98

greptimedb

GreptimeTeam/greptimedb · ★ 6.7k ▲23/7d · Rust

GreptimeDB is an observability database that stores metrics, logs, and traces in one columnar table model, with object storage holding the durable data. It accepts familiar telemetry protocols and lets operators query with SQL or PromQL instead of maintaining a separate database for each signal.

SetupDocsCommunityMaturity

Verdict Our 3-CPU sandbox installed 1,579 packages in 207 seconds, then both build and tests hit 900-second caps, so GreptimeDB is a serious systems project with a high source-work cost. Try the published standalone image if one database for metrics, logs, and traces would simplify your operation.

Full review →

99

tchMaterial-parser

happycola233/tchMaterial-parser · ★ 6.6k ▲78/7d · Python

tchMaterial-parser is a Chinese-language desktop app for finding and downloading electronic textbooks from China's National Smart Education Platform. Its README and user instructions are in Chinese, and the repository provides no English documentation; the app can batch URLs, name PDFs, and optionally add bookmarks.

SetupDocsCommunityMaturity

Verdict Our tchMaterial-parser run installed 39 packages, built in 4 seconds, and passed its tests in 5 seconds, so version 4.2 is an easy technical trial for Chinese-speaking desktop users. Use it for occasional, permitted textbook downloads when the official site's manual flow is too slow.

Full review →

100

cli

brightdata/cli · ★ 6.3k · TypeScript

Bright Data CLI puts the company's scraping, search, hosted browser, and structured dataset APIs behind one terminal command. It solves the glue-work problem for developers who already want Bright Data services and need output that can feed shell scripts or coding agents.

SetupDocsCommunityMaturity

Verdict Our Bright Data CLI run installed 88 packages in 14 seconds, then passed its build and tests, so trying the client is cheap even though using it ties the workflow to Bright Data's remote services. Choose it when your account already pays for those services and one scriptable interface is worth more than provider independence.

Full review →

101

stdlib

stdlib-js/stdlib · ★ 6.0k ▲40/7d · JavaScript

stdlib is a large standard library for numerical and scientific computing in JavaScript, TypeScript, and some C. It supplies focused packages for math, statistics, arrays, random numbers, data processing, streams, plotting, and related utilities so developers do not have to assemble those basics from unrelated modules.

SetupDocsCommunityMaturity

Verdict stdlib is the best fit for JavaScript teams that care about numerical breadth, typed APIs, and the option to install one exact function at a time. Do not begin with the full repository package unless you need the REPL or a wide part of the library: our install was huge, its audit was noisy, and its test run hit an expired certificate.

Full review →

102

os-taxonomy

withmarbleapp/os-taxonomy · ★ 4.4k · JavaScript

Marble Skill Taxonomy is a set of JSON files that breaks primary and elementary learning into 1,590 small topics connected by 3,221 prerequisite links. It gives education products a starting map for deciding what a child may need to learn before moving to another concept.

SetupDocsCommunityMaturity

Verdict Our os-taxonomy run installed 0 packages in 10 seconds, but the repository exposed no build or test target for us to run. Use it as a well-labeled seed dataset if your team can review the graph and handle its layered licenses.

Full review →

103

webclaw

0xMassi/webclaw · ★ 2.3k ▲7/7d · Rust

Webclaw is a Rust web extraction toolkit that turns pages into Markdown, plain text, JSON, cleaned HTML, or compact context for language models. It runs as a CLI, MCP server, library, or self-hosted API, with a separate paid service for rendering, protected sites, search, and managed jobs.

SetupDocsCommunityMaturity

Verdict Our Webclaw run installed 384 packages, built in 171 seconds, and passed all 772 tests in 105 seconds, an unusually clean result for a web extraction project. Use it for local public-page extraction and MCP access when Rust binaries, multiple output formats, and a small checkout appeal.

Full review →

104

indicator

cinar/indicator · ★ 1.8k ▲81/7d · Go

Indicator is a Go library with more than 80 technical-analysis indicators, trading strategies, data repositories, and a backtesting framework. Version 2 sends data through Go channels, while companion commands can fetch Tiingo data, generate HTML reports, or expose selected strategies through a Model Context Protocol server.

SetupDocsCommunityMaturity

Verdict Our Indicator build took 26 seconds, and all 30 reported tests passed in 13 seconds. Use it for Go-based research when channel streams, AGPL-3.0, and hands-on validation fit your team.

Full review →

105

supabase

supabase/supabase · ★ 109k ▲357/7d · TypeScript

Supabase is a Postgres-based backend platform that bundles a database, authentication, generated APIs, file storage, realtime updates, functions, and a management dashboard. It lets application teams start with managed infrastructure or run the same broad stack themselves instead of assembling every backend service separately.

SetupDocsCommunityMaturity

Verdict Our Supabase install pulled 3,094 packages and used 2,934 MB, then the build failed at the federated-content task after 36 seconds. The managed service remains a strong choice for teams that want Postgres plus application services, because they do not have to build this monorepo to use it.

Full review →

106

superset

apache/superset · ★ 75k ▲81/7d · Python

Apache Superset is a self-hosted business intelligence application for exploring SQL data, writing queries, building charts, and sharing dashboards. It gives analysts a visual workspace while administrators connect database engines, define access, and optionally expose an MCP server to AI clients.

SetupDocsCommunityMaturity

Verdict Our Superset run installed 162 packages in 88 seconds and built in 13 seconds, but pytest stopped after 200 collection/setup errors because SQLite lacked the `css_templates` table. Superset is worth the operating cost for a staffed data platform that needs broad SQL connectivity, detailed permissions, and several ways to publish analysis.

Full review →

107

nocodb

nocodb/nocodb · ★ 65k ▲78/7d · TypeScript

NocoDB puts a spreadsheet-like interface on its own data store or connected SQL databases, with grid, form, gallery, Kanban, and calendar views. It gives non-database users a place to edit records while developers retain APIs, access controls, automations, and self-hosting.

SetupDocsCommunityMaturity

Verdict Our NocoDB install consumed 3,019 MB across 3,948 packages, then exposed no root build or test target for the harness, so source adoption needs more scrutiny than the one-command Docker demo suggests. Use it when several people genuinely need spreadsheet views, permissions, APIs, and workflows around shared data.

Full review →

108

ray

ray-project/ray · ★ 44k ▲68/7d · Python

Ray is a Python framework for running work across multiple CPU or GPU processes, machines, and clusters. It gives developers one runtime for distributed tasks, stateful actors, shared objects, and higher-level libraries for data processing, training, tuning, reinforcement learning, and serving.

SetupDocsCommunityMaturity

Verdict Our source build failed in 13 seconds, and pytest reached 200 collection or setup errors before running a single test. That does not erase Ray's broad, coherent distributed-computing design, but it does mean teams should evaluate the packaged installation separately from contributor setup.

Full review →

109

surrealdb

surrealdb/surrealdb · ★ 33k ▲37/7d · Rust

SurrealDB is a Rust database that stores relational, document, graph, time-series, geospatial, full-text, and vector data behind its own SQL-like language. It aims to replace several specialized stores and some API-layer code with one server or embedded engine, including live queries and row-level permissions.

SetupDocsCommunityMaturity

Verdict Our SurrealDB checkout spent 900 seconds on the build and another 900 seconds on tests without finishing compilation, so source validation is expensive on a 3-CPU box. The published binary is still worth a prototype when graph links, live queries, permissions, and search must share one data model.

Full review →

110

influxdb

influxdata/influxdb · ★ 32k ▲15/7d · Rust

InfluxDB 3 Core is a database for ingesting, storing, and querying time-stamped events such as metrics, sensor readings, and market data. The main branch stores Parquet on local disk or object storage and exposes SQL, InfluxQL, FlightSQL, line protocol, and HTTP interfaces.

SetupDocsCommunityMaturity

Verdict Our InfluxDB 3 Core build spent 567 seconds compiling before `libpython3.11` stopped both build and tests, so source adopters should provision the documented Python and protobuf dependencies first. Prefer official packages for an evaluation unless contributing to the Rust codebase.

Full review →

111

anki

ankitects/anki · ★ 31k ▲168/7d · Rust

Anki is the source code for a desktop flashcard program that schedules reviews according to how well you remember each card. It stores structured notes, renders card templates with media, and can synchronize collections through AnkiWeb so study can continue across supported clients.

SetupDocsCommunityMaturity

Verdict Our Anki checkout pulled 688 Yarn packages and 507 MB in 37 seconds, then its web build failed after 4 seconds on missing generated paths. Use the packaged app if your goal is studying; use this repository only when an add-on cannot do the job and you can work through its multi-language build and review rules.

Full review →

112

label-studio

HumanSignal/label-studio · ★ 28k ▲37/7d · TypeScript

Label Studio is a self-hosted tool for having people label audio, text, images, video, and time-series data. It turns raw examples into structured training data and can export annotations to model-oriented formats. Its configurable interface makes one system useful across several kinds of labeling work.

SetupDocsCommunityMaturity

Verdict Our build succeeded in 6 seconds, but pytest stopped with 44 collection/setup errors before running a single test. Label Studio is still a strong default when one self-hosted system must cover several annotation types, especially for teams able to own its deployment and validation.

Full review →

113

dash

plotly/dash · ★ 24k ▲8/7d · Python

Dash is a Python framework for building interactive data apps that run in a browser. It connects controls such as dropdowns and sliders to Python callbacks and Plotly.js charts, so a data team can make a working interface without starting with a React application.

SetupDocsCommunityMaturity

Verdict Our Dash run installed 775 packages in 25 seconds, but both the 7-second build and 7-second test step failed because required commands were absent. Dash is still a sensible choice for Python teams that want Plotly charts and explicit reactive callbacks in one codebase.

Full review →

114

airbyte

airbytehq/airbyte · ★ 22k ▲56/7d · Python

Airbyte is a data-movement platform that copies records and files from APIs, databases, and SaaS tools into warehouses, lakes, databases, and AI systems. It provides a large connector catalog, managed Cloud service, self-hosted platform, and tools for building or modifying connectors.

SetupDocsCommunityMaturity

Verdict Our Airbyte documentation install pulled 1,892 packages, and its production build was killed after 179 seconds, so even the docs slice carries monorepo-scale weight. Airbyte is a sensible default evaluation when connector breadth saves more engineering time than the platform costs to run and verify.

Full review →

115

kibana

elastic/kibana · ★ 21k ▲5/7d · TypeScript

Kibana is the web interface for querying, exploring, visualizing, and managing data held in Elasticsearch. It supplies dashboards and operational applications for search, observability, and security, with a hosted Elastic Cloud option for teams that do not want to run the stack themselves.

SetupDocsCommunityMaturity

Verdict Our Kibana install took 761 seconds and 3,122 MB, then the distributable build failed after 38 seconds, so building this source tree is specialist work even before Elasticsearch enters the picture. Use Kibana when Elasticsearch is already your data platform and its first-party applications justify synchronized upgrades.

Full review →

116

IPTV

Free-TV/IPTV · ★ 21k ▲165/7d · Python

Free-TV/IPTV is an English-documented M3U playlist that points compatible players at freely available television streams from many countries and online services. It replaces the work of collecting channel URLs, logos, groups, country codes, and electronic program guide identifiers by hand.

SetupDocsCommunityMaturity

Verdict Our Free-TV/IPTV run installed 35 packages in 22 seconds and built in 27 seconds, but it skipped tests and did not play any of the 2,053 stream entries. Use the playlist as a convenient, disposable input for a player, especially when country grouping and EPG identifiers save setup time.

Full review →

117

bokeh

bokeh/bokeh · ★ 20k · Python

Bokeh is a Python library for turning data into interactive charts, dashboards, and browser-based applications. It is meant for people who want web-ready visualization without building the entire presentation layer in JavaScript.

SetupDocsCommunityMaturity

Verdict Our install completed in 52 seconds, but both the 17-second build and 8-second test command exited 1. Bokeh remains a strong choice when a Python team needs genuinely interactive browser graphics and wants a mature, permissively licensed project behind them.

Full review →

118

dbx

t8y2/dbx · ★ 20k ▲1.2k/7d · Rust

DBX is a database manager whose default README is in English, with a Chinese translation; its release notes and much of its issue discussion are Chinese. It covers SQL databases, document and key-value stores, search systems, message queues, and service registries through desktop, web, Docker, CLI, AI-assistant, and MCP interfaces.

SetupDocsCommunityMaturity

Verdict Our DBX run built in 16 seconds and passed 10,947 of 10,950 tests, but 2 failures kept the 746-second suite red. It is a useful read-oriented client for developers who cross database families and want controlled MCP access.

Full review →

119

txtai

neuml/txtai · ★ 13k ▲9/7d · Python

txtai is a Python framework that combines vector search, relational queries, graphs, model pipelines, workflows, and agents around one embeddings database. It helps developers build local semantic search and retrieval applications without stitching together a separate vector store, model server, workflow engine, and API for every prototype.

SetupDocsCommunityMaturity

Verdict Our txtai install took 369 seconds, consumed 6,602 MB, and built successfully, but pytest collected 0 tests and pip-audit found 1 known vulnerability. It remains a good fit for a Python team that genuinely wants search, models, and workflows in one process.

Full review →

120

liteparse

run-llama/liteparse · ★ 12k ▲36/7d · Rust

LiteParse is a local document parser built in Rust, with bindings for Node.js, Python, Rust, and browsers. It extracts spatial text, bounding boxes, Markdown, images, forms, annotations, and page screenshots from PDFs, and can convert office documents before parsing them.

SetupDocsCommunityMaturity

Verdict Our LiteParse source build spent 378 seconds compiling before a Tesseract language-data download timed out, and the test step failed on the same download path. The project is still a serious candidate for local document pipelines because its published packages cover 4 runtimes and its API exposes useful spatial detail.

Full review →

121

quickwit

quickwit-oss/quickwit · ★ 12k ▲43/7d · Rust

Quickwit is a distributed search engine that keeps log and trace indexes in cloud object storage while search and indexing workers scale separately. It gives observability teams native APIs plus partial Elasticsearch compatibility, which can reduce the amount of indexed data held on attached server disks.

SetupDocsCommunityMaturity

Verdict Our Quickwit build took 560 seconds and 2 of 2,178 tests failed, so its object-storage design deserves a staged trial rather than a blind Elasticsearch replacement. It is worth testing for long-retention logs or traces when the team already operates object storage, PostgreSQL, and Kafka.

Full review →

122

awesome-zhuiju-free

laoma2053/awesome-zhuiju-free · ★ 8.6k ▲320/7d · JavaScript

Awesome Zhuiju Free is a Chinese-language directory of free streaming sites, media apps, cloud-drive search, BitTorrent indexes, subtitles, TVBox configurations, IPTV sources, and related open-source projects. The repository has no English README, so English-only readers need translation to use its ratings, risk notes, and contribution process.

SetupDocsCommunityMaturity

Verdict Our Awesome Zhuiju Free run installed 35 packages in 37 seconds and built in 8 seconds, but the repository supplied no test target. Use its 114-entry Chinese directory as a lead sheet, then check every destination's rights, ads, privacy, and playback yourself.

Full review →

123

awesome-free-apps

Axorax/awesome-free-apps · ★ 7.7k ▲44/7d · JavaScript

Awesome Free Apps is a large, human-curated directory of free desktop and mobile software. It groups tools by job, marks supported operating systems, identifies open-source entries, and offers generated filters for people who do not want to search one enormous list.

SetupDocsCommunityMaturity

Verdict Awesome Free Apps is a very good browsing shelf and a poor substitute for product research. Its breadth, platform markers, and generated filters make discovery fast, while its thin entries and live-link maintenance risk put verification on the reader.

Full review →

124

dbgate

dbgate/dbgate · ★ 7.3k ▲12/7d · JavaScript

DbGate is a desktop and browser-based database manager for relational databases, MongoDB, Redis, and several other engines. It gives developers one place to inspect data, edit schemas, write queries, compare structures, and move records between common file formats.

SetupDocsCommunityMaturity

Verdict DbGate is a good fit for developers who want one familiar interface across SQL, document, and key-value databases, especially when both desktop and self-hosted web access matter. The source tree is heavier and less uniform than the two-line development start suggests, and current reports around paged results and RDS IAM deserve a trial with your own database.

Full review →

125

lance

lance-format/lance · ★ 7.1k ▲31/7d · Rust

Lance is a columnar file and table format for datasets that mix embeddings with text, images, audio, or video. It gives data teams versioned tables, random row access, vector and full-text indexes, and SQL-style filtering without requiring a separate vector store for each dataset.

SetupDocsCommunityMaturity

Verdict Our Lance build completed in 715 seconds, but its tests were still compiling when our 900-second limit expired, so adopting the source tree demands patience even before data validation starts. Use Lance when embeddings and source media belong in the same object-storage table and random access is central to the workload.

Full review →

126

FalkorDB

FalkorDB/FalkorDB · ★ 6.1k ▲154/7d · Rust

FalkorDB is a graph database that stores connected data as nodes and relationships and queries it with OpenCypher. It runs as a Redis module, uses sparse matrices for graph operations, and targets knowledge graphs, fraud analysis, security data, and application memory.

SetupDocsCommunityMaturity

Verdict Our FalkorDB build failed after 116 seconds and its tests failed after 7 seconds because `deps/RediSearch/bin` was missing. Use the published image for a product trial, and use the recommended development container if you plan to contribute.

Full review →

127

google-maps-scraper

gosom/google-maps-scraper · ★ 5.9k ▲88/7d · Go

Google Maps Scraper is a Go application for collecting public business listings, including names, phone numbers, websites, ratings, reviews, coordinates, and optional email addresses. It turns repetitive Maps research into CSV or JSON output and can also serve that workflow through a browser interface, REST API, or self-hosted platform.

SetupDocsCommunityMaturity

Verdict Our build completed in 72 seconds, and all 28 tests passed. Google Maps Scraper is one of the more convincing self-hosted choices for turning Maps searches into operational data because it offers a CLI, Web UI, REST API, and a path toward distributed jobs.

Full review →

128

opentelemetry-collector-contrib

open-telemetry/opentelemetry-collector-contrib · ★ 4.9k ▲22/7d · Go

OpenTelemetry Collector Contrib is the large community repository of receivers, processors, exporters, connectors, and extensions for the OpenTelemetry Collector. It lets operators ingest telemetry from many systems, transform it, and send traces, metrics, or logs to different backends without putting vendor-specific code in every application.

SetupDocsCommunityMaturity

Verdict Our root test command finished in 6 seconds with 0 tests because it matched no Go packages, so this repository must be evaluated through its multi-module Make targets rather than `go test ./...`. Use Contrib when a specific receiver, processor, or exporter justifies its stability and support level.

Full review →

129

martin

maplibre/martin · ★ 3.9k ▲18/7d · Rust

Martin is a tile server that turns PostGIS tables and GeoJSON into vector tiles, or serves existing PMTiles and MBTiles archives. It also handles styles, sprites, fonts, composite sources, bulk tile generation, and MBTiles inspection without requiring a separate map application server.

SetupDocsCommunityMaturity

Verdict Martin is an excellent fit when one deployment must cover live PostGIS tiles and archive formats while retaining serious MBTiles tooling. Use a release binary or official container unless you are ready to supply native build dependencies, because our clean source build did not complete.

Full review →

130

Ontology-Playground

microsoft/Ontology-Playground · ★ 2.7k ▲75/7d · TypeScript

Microsoft Ontology Playground is a preview web application for learning ontology concepts, drawing entity-and-relationship models, and exporting RDF/XML for Microsoft Fabric IQ. The default site is fully static, so a class, teacher, or analyst can explore examples and edit diagrams without operating a database or application server.

SetupDocsCommunityMaturity

Verdict Our Ontology Playground run installed 314 packages in 19 seconds and passed all 397 tests, but npm audit found 11 known vulnerabilities, including 6 high-severity findings. It is a good teaching and early-design tool for small Fabric IQ models because the static app combines examples, a visual editor, RDF import and export, and lessons.

Full review →

131

flights

AWeirdDev/flights · ★ 2.1k ▲88/7d · Python

fast-flights is a Python library that builds Google Flights searches and parses the returned JavaScript data into typed results. It gives developers flight search data without an official Google Flights API, but it depends on an undocumented web interface that Google can change.

SetupDocsCommunityMaturity

Verdict Our fast-flights run installed 37 packages and passed all 8 tests in 35 seconds across install, build, and test steps. It is a good fit for prototypes and internal fare checks where a missed result can be caught, retried, or compared with the website.

Full review →

132

crawl4ai

unclecode/crawl4ai · ★ 84k ▲1.6k/7d · Python

Crawl4AI is a Python browser crawler that turns dynamic web pages into Markdown, structured data, screenshots, links, and other artifacts for RAG, agents, and data pipelines. It gives developers direct control over browser sessions, JavaScript, proxies, extraction rules, deep crawls, and an optional Docker API instead of requiring a hosted scraping account.

SetupDocsCommunityMaturity

Verdict Our Crawl4AI install consumed 5,368 MB, then the build failed in 12 seconds and pytest ended with 35 collection or setup errors, so this commit did not provide a clean contributor path in our sandbox. Use the packaged release for browser-heavy Python crawling only after a real-site pilot and container soak test.

Full review →

133

MinerU

opendatalab/MinerU · ★ 80k ▲491/7d · Python

MinerU is a Python document-extraction project that turns difficult source documents into content that downstream tools can work with. It is aimed at teams that need an open-source extraction layer instead of manually copying material or depending entirely on a hosted converter.

SetupDocsCommunityMaturity

Verdict Our build succeeded in 7 seconds, but the test run ended with 8 passes and 1 collection/setup error, so MinerU is promising rather than drop-in dependable. Its active development, recent 3.4.5 release, multiple demos, and unusually large community make it worth a serious pilot for document-heavy systems.

Full review →

134

orm

prisma/orm · ★ 48k · TypeScript

Prisma ORM is a TypeScript toolkit for defining how an application talks to its database and for generating the code that performs those reads and writes. The Prisma 8 release candidate focuses on PostgreSQL and MongoDB, with project scaffolding, reusable agent instructions, and an extension system intended to reduce custom database plumbing.

SetupDocsCommunityMaturity

Verdict Our build completed in 130 seconds, but the test command reached the 900-second cap with 108 passed, 0 failed, and 16 skipped. Prisma 8 looks unusually deliberate about agent-guided database work and extension authorship, yet it is still a release candidate with a narrower database list than many established ORMs.

Full review →

135

zju-icicles

QSCTech/zju-icicles · ★ 41k ▲35/7d · HTML

QSCTech/zju-icicles is a Chinese-language, crowd-sourced archive of course materials for Zhejiang University, mainly computer science. The README and most repository content are in Chinese, and there is no English documentation. It solves the scattered, word-of-mouth problem of university course intelligence: past exams, notes, homework answers, A4 cheat sheets, and course reviews become a browsable GitHub repo.

SetupDocsCommunityMaturity

Verdict Our run cloned 6,979.5 MB and installed 58 Python packages in 25 seconds, then a generic build failed in 5 seconds, but that failure is irrelevant because this repo is a document archive, not software. If you are a ZJU student, use it without hesitation; if you are not, steal its directory-and-README model for your own university..

Full review →

136

drawdb

drawdb-io/drawdb · ★ 40k ▲157/7d · JavaScript

drawDB is a browser-based database diagram editor that can turn schemas into visual tables and relationships, then import or export SQL. It works without an account for individual diagrams, while optional sharing depends on a separate server project.

SetupDocsCommunityMaturity

Verdict Our drawDB build finished in 7 seconds with 0 audit findings, but installing its 1,890 packages used 618 MB and no test target checked SQL round trips. Use it for quick schema design and reviewed SQL export, especially when an account-free browser canvas is the priority.

Full review →

137

timesfm

google-research/timesfm · ★ 33k ▲492/7d · Python

TimesFM is Google's pretrained model for predicting future values of a time series without first training a model on that particular series. The open Python package returns point and quantile forecasts, with optional covariates for known influences on the target.

SetupDocsCommunityMaturity

Verdict Our TimesFM run passed 26 tests with 0 assertion failures, but 6 modules could not collect because `pandas` was missing and the audit found 5 known vulnerabilities. Put it in a forecasting bake-off when a zero-shot point and quantile baseline could save training work.

Full review →

138

developer-portfolios

emmabostian/developer-portfolios · ★ 27k ▲85/7d · Python

Developer Portfolios is an alphabetical directory of personal websites submitted by developers, designers, students, and technical specialists. It gives people a large pool of real portfolio examples to browse and exposes the entries as Markdown plus a generated JSON feed.

SetupDocsCommunityMaturity

Verdict Our Developer Portfolios run built in 4 seconds and passed 38 of 39 tests, while pip-audit found 0 known vulnerabilities, so the repository is easy to inspect but has one concrete normalization defect. Use it as a broad browsing shelf, then shortlist examples that match your role and verify each link yourself.

Full review →

139

SpacetimeDB

clockworklabs/SpacetimeDB · ★ 25k ▲68/7d · Rust

SpacetimeDB is a source-available relational database that also runs application logic and sends live changes to connected clients. Developers put tables and server functions in Rust, C#, TypeScript, or C++ modules, removing the usual API server from many real-time apps and multiplayer games.

SetupDocsCommunityMaturity

Verdict Our SpacetimeDB checkout installed 2,801 packages and used 1,602 MB, then failed both its 6-second build and its 15-second test run, so source contributors should expect repair work before a clean first pass. The product is still worth a prototype for multiplayer or synchronized applications because its database, server logic, and subscriptions form one understandable model.

Full review →

140

neon

neondatabase/neon · ★ 23k ▲29/7d · Rust

Neon is an English-language, open-source Postgres platform that separates stateless compute nodes from a distributed storage layer. That split supports database branches and compute that can scale independently, while applications still connect with ordinary Postgres clients and drivers.

SetupDocsCommunityMaturity

Verdict Our Python-side build passed in 8 seconds, but tests failed 10 seconds later because pytest could not import `allure`. Neon is a strong hosted choice for teams that want Postgres compatibility, branching, and independent compute without operating the storage system themselves.

Full review →

141

pdf-inspector

firecrawl/pdf-inspector · ★ 19k ▲180/7d · Rust

PDF Inspector is a Rust parser that classifies PDFs, extracts positioned text, and converts native text into Markdown. It has Rust, command-line, Python, Node.js, and browser packages, with selective OCR available on native paths for pages that lack usable text.

SetupDocsCommunityMaturity

Verdict Our PDF Inspector run passed 100 of 101 tests, with the lone failure raising an encrypted-PDF error after a 474-second install. It is a strong candidate for local native-text extraction and selective OCR routing, provided you test the fonts and tables that matter to your corpus.

Full review →

142

open-source-games

bobeff/open-source-games · ★ 14k · Python

open-source-games is a curated README of playable open-source games, source ports, decompilations, and remakes of commercial titles. It gives developers and players a genre-based starting point with links to project sites and source locations.

SetupDocsCommunityMaturity

Verdict Our run built the 4-file open-source-games repository in 5 seconds, but there was no test target and nothing checked its 183 source-labeled links. Use it as a quick genre index when you are willing to vet every candidate's license, assets, platform, and current activity yourself.

Full review →

143

lancedb

lancedb/lancedb · ★ 11k ▲48/7d · Rust

LanceDB is an embedded retrieval database for vectors, text, metadata, and multimodal files, built on the Lance columnar format. Python, TypeScript, and Rust applications can keep local tables in process, while the same project also connects to a managed remote service for teams that do not want to operate storage themselves.

SetupDocsCommunityMaturity

Verdict Our LanceDB build ran for 806 seconds and failed because `-lpython3.11` was missing, then the test command hit the same linker error after 758 seconds, so this source checkout did not clear either gate. Try a published SDK when you want an embedded multimodal retrieval layer and value Arrow ecosystem integration.

Full review →

144

check-if-email-exists

reacherhq/check-if-email-exists · ★ 9.9k ▲327/7d · Rust

check-if-email-exists is a Rust library, command-line tool, and HTTP backend that examines an email address without sending a message. It checks syntax, DNS mail records, disposable and role accounts, SMTP responses, catch-all behavior, disabled or full mailboxes, and returns a cautious reachability category.

SetupDocsCommunityMaturity

Verdict Our run built in 311 seconds and passed all 33 Rust tests, so the code is a credible base for teams that understand SMTP verification. Use it for a risk signal, not as proof that a person owns an address or that every future message will arrive.

Full review →

145

chinese-poetry-api

palemoky/chinese-poetry-api · ★ 2.8k ▲32/7d · Go

Chinese Poetry API is a Go service that exposes a large classical Chinese poetry corpus through REST and GraphQL. Its README and operator documentation are written in Chinese, and we found no English guide; the API gives applications searchable poem, author, dynasty, and form data in simplified or traditional Chinese.

SetupDocsCommunityMaturity

Verdict Our Chinese Poetry API build passed in 66 seconds, but 1 of 2 test groups failed because the SQLite runtime lacked FTS5. Use the published container when you need a self-hosted Chinese poetry search service and can verify the source text your application displays.

Full review →

146

Legendary_OSINT

K2SOsint/Legendary_OSINT · ★ 1.7k ▲8/7d

This is a curated collection of links to Open-Source Intelligence (OSINT) tools and resources. It solves the discovery problem for investigators and researchers by organizing hundreds of web-based tools, databases, and software into 25 distinct, easy-to-navigate categories.

SetupDocsCommunityMaturity

Verdict Legendary_OSINT is a well-organized and highly useful directory for discovering intelligence-gathering tools. While not as exhaustive as older lists, its clean structure and modern categories make it a valuable bookmark.

Full review →

147

ai-data-extraction

0xSero/ai-data-extraction · ★ 1.3k ▲4/7d · Python

AI Data Extraction is a set of local Python scripts that reads stored conversations from coding assistants and converts them into a common JSONL shape. It targets Claude Code, Codex, Cursor, Trae, Windsurf, Continue, Gemini CLI, and OpenCode, including prompts, replies, file context, edits, tool calls, and metadata when the source format contains them.

SetupDocsCommunityMaturity

Verdict Our sandbox installed 35 packages and passed all 6 tests in 19 seconds combined, so the code is cheap to inspect and trial. Use it only on data you own or are authorized to process, behind a separate redaction and review step, because the reviewed branch writes sensitive coding history without the protections proposed in two open pull requests.

Full review →

148

shulihuazixuecongshu

tradecatlabs/shulihuazixuecongshu · ★ 897 · Python

Shulihuazixuecongshu is a Chinese-language preservation and ebook-reconstruction project for a 17-volume mathematics, physics, and chemistry self-study series. Its documentation is Chinese, with no English guide present. It keeps scans, normalized Markdown, images, and scripts that build reproducible EPUB 3 files.

SetupDocsCommunityMaturity

Verdict Our run installed 35 packages in 14 seconds and completed the build in 7 seconds, but the repository supplied no test target. Use it for Chinese-language study or auditable ebook preservation when you can verify the source corrections and the rights for your jurisdiction.

Full review →

149

MediaCrawler

NanmiCoder/MediaCrawler · ★ 65k ▲318/7d · Python

MediaCrawler is a Chinese-first social media collector, and a separate English README is available. It uses a logged-in browser session to gather public posts, creator pages, comments, and media from Xiaohongshu, Douyin, Kuaishou, Bilibili, Weibo, Tieba, and Zhihu.

SetupDocsCommunityMaturity

Verdict Our MediaCrawler checkout installed 269 npm packages in 26 seconds, but it exposed no npm build or test target for us to verify. Use it as learning material or for a narrow, permitted research collection where breakage can be handled by a developer.

Full review →

150

qlib

microsoft/qlib · ★ 49k ▲168/7d · Python

Qlib is a Python research platform for testing machine-learning ideas on financial-market data, then turning predictions into portfolios and backtests. It gives quant teams one place for data preparation, model training, experiment records, analysis, and simulated execution.

SetupDocsCommunityMaturity

Verdict Our Qlib run installed 214 packages and built successfully; 10 tests failed and 9 more ended in collection or setup errors, so it suits quant research teams that can validate the stack themselves. Use it when one framework for data, models, experiments, and backtests is worth a 964 MB environment and substantial method checking.

Full review →

151

turso

tursodatabase/turso · ★ 24k ▲79/7d · Rust

Turso is an in-process SQL database written in Rust that aims to work with SQLite files, SQL, and APIs while adding features such as concurrent writes and change tracking. It also has an experimental Postgres-compatible frontend, letting one database core understand more than one SQL dialect.

SetupDocsCommunityMaturity

Verdict Our build succeeded in 5 seconds, but the test run ended with 8 collection or setup errors. Turso is one of the more ambitious SQLite-adjacent projects, and its active development, broad bindings, MVCC, and CDC make it worth a serious evaluation.

Full review →

152

memvid

memvid/memvid · ★ 17k ▲10/7d · Rust

Memvid is an embeddable Rust memory store that keeps content, metadata, a write-ahead log, and search indexes inside one `.mv2` file. It is meant to give agents portable lexical, vector, and time-aware recall without running a separate database server.

SetupDocsCommunityMaturity

Verdict Our Memvid build finished in 105 seconds, but the full test run hit the 900-second cap with 3 lexical tests still running beyond 60 seconds. Try it for single-writer, offline memory where copying one `.mv2` file is genuinely useful.

Full review →

153

prettymaps

marceloprates/prettymaps · ★ 14k ▲30/7d · Python

Prettymaps is a Python library that turns OpenStreetMap data into stylized map images. Give it a place name, coordinates, or a geographic boundary, then adjust the map through presets, layer queries, colors, and normal Matplotlib objects.

SetupDocsCommunityMaturity

Verdict Our Prettymaps run installed 148 packages and used 1,913 MB, then built in 6 seconds and passed all 9 collected tests. That result supports trying it for inspected, one-off map artwork, especially when its presets and exposed GeoDataFrames save you from assembling the plotting stack yourself.

Full review →

154

czsc

waditu/czsc · ★ 6.2k ▲165/7d · Rust

CZSC is a Chinese-first Python and Rust toolkit for applying Chan theory to market charts, building signals, and researching trading strategies. No official English documentation is included in the repository, although the README links to a generated DeepWiki reference. It turns candlestick data into Chan structures, multi-timeframe signals, positions, visualizations, and backtest inputs.

SetupDocsCommunityMaturity

Verdict Our CZSC run installed 57 packages in 753 seconds and passed 516 of 517 executed tests, with one datetime-precision assertion failing. Choose it when Chan theory and Chinese documentation already fit the research team.

Full review →

155

pgrust

malisper/pgrust · ★ 5.0k ▲65/7d · Rust

pgrust is an experimental rewrite of the PostgreSQL 18.3 server in Rust that aims to keep its wire protocol and SQL behavior. It replaces major internals with threaded concurrency, a vectorized executor, JIT compilation, query scheduling, and a columnar layout, while openly warning that it is not ready for production data.

SetupDocsCommunityMaturity

Verdict Our pgrust build passed in 443 seconds and 152 of 153 tests passed, with the remaining case failing because `en_US.UTF-8` was absent from our Debian container. That is a good research result and nowhere near permission to store valuable data, a boundary the README states plainly.

Full review →

156

iggy

apache/iggy · ★ 4.9k ▲49/7d · Rust

Apache Iggy is a persistent message-streaming server written mainly in Rust, with TCP, QUIC, WebSocket, and optional HTTP interfaces. It stores ordered messages in streams, topics, and partitions for systems that need a self-hosted event log rather than a managed broker.

SetupDocsCommunityMaturity

Verdict Our 10-second Iggy run installed only 2 npm packages because the root package had no build or test target, so it says nothing about whether the 611,453-line Rust server compiles or passes its suite. Iggy deserves a workload trial for teams attracted to its single-binary server and multi-protocol clients, but that trial must use the Cargo or Docker path.

Full review →

157

RuVector

ruvnet/RuVector · ★ 4.5k ▲18/7d · Rust

RuVector is a Rust-based vector store and memory toolkit for AI agents that need to recall information across sessions. It combines local embeddings, persistent similarity search, graphs, feedback-driven learning, snapshots, and optional shared memory, with entry points for Rust, Node.js, browsers, PostgreSQL, HTTP, and MCP.

SetupDocsCommunityMaturity

Verdict RuVector is worth testing for an embedded agent-memory system when you can start with `VectorDB` or the hooks CLI and ignore most of the monorepo. Its documentation is frank about incomplete paths, but the failed root scripts and 33 audit findings make a broad production adoption hard to recommend today.

Full review →

158

msgvault

kenn-io/msgvault · ★ 2.1k ▲4/7d · Go

msgvault is a local archive for email, chat, calendar events, meeting notes, contacts, and attachments. It syncs or imports data into SQLite and content-addressed storage, then provides offline search, DuckDB analytics, a Web UI, a TUI, and MCP access.

SetupDocsCommunityMaturity

Verdict Our msgvault build took 128 seconds, and the test run timed out at 900 seconds with 34 passed and 6 failed packages reported. It is one of the more ambitious personal-message archives, especially for people who need several sources, attachments, local analytics, and MCP in one place.

Full review →

159

harvey-labs

harveyai/harvey-labs · ★ 1.4k ▲22/7d · Python

Harvey LAB is a dataset and execution harness for testing whether AI agents can complete realistic legal assignments. It gives an agent synthetic matter documents, collects editable deliverables, and uses criterion-by-criterion model judges to score work such as diligence reviews, drafting, extraction, and legal research.

SetupDocsCommunityMaturity

Verdict Our Harvey LAB run passed 12,588 tests but still ended with 23 failures and 39 collection or setup errors, so this benchmark deserves inspection before its scores reach a procurement slide. Use it to compare legal agents when lawyers can audit the task contracts, source documents, and judge reasoning.

Full review →

160

crw

us/crw · ★ 1.0k ▲64/7d · Rust

fastCRW is a Rust service and command-line tool that turns web pages into markdown, links, screenshots, or structured JSON. It combines scraping, crawling, URL discovery, web search, and schema extraction behind its own API, a Firecrawl-compatible route, SDKs, and an MCP server for coding agents.

SetupDocsCommunityMaturity

Verdict fastCRW is a credible choice when one compact service must feed web data to applications and MCP agents, particularly for teams testing a move from Firecrawl. The local option buys control, but it also hands you browser rendering, proxy quality, search configuration, and licensing decisions.

Full review →

161

ai-data-extractor

bawadou/ai-data-extractor · ★ 555 ▲3/7d · Python

AI Data Extractor is a Python command-line tool that copies local coding-assistant conversations into a common JSONL format. It covers ten sources, including Claude Code, Codex CLI, Cursor, Windsurf, Cline, Aider, and Gemini CLI, so developers can back up or inspect histories that are otherwise scattered across app folders.

SetupDocsCommunityMaturity

Verdict Our run installed 35 packages in 21 seconds and found 0 known vulnerabilities, but this 1,707-line extractor has no automated test target. Use it when the alternative is manually spelunking through ten unrelated storage layouts, and treat every generated file as sensitive.

Full review →

162

ChinaTextbook

TapXWorld/ChinaTextbook · ★ 82k ▲274/7d · Roff

ChinaTextbook is a massive GitHub repository that collects and provides free access to Chinese educational textbooks as PDFs. The project and its materials are entirely in Chinese, created to combat the reselling of free public resources and to help overseas Chinese families access the mainland curriculum.

SetupDocsCommunityMaturity

Verdict ChinaTextbook is an immensely popular and valuable archive of Chinese educational materials, but it's best viewed as a dormant library, not an active project. For its target audience, it's a wonderfully convenient way to get a complete set of math textbooks.

Full review →

163

Kronos

shiyu-coder/Kronos · ★ 39k ▲217/7d · Python

Kronos is a family of Python models built to forecast financial candlestick sequences such as open, high, low, close, and volume data. It tokenizes continuous market values and feeds those tokens to an autoregressive Transformer, with published checkpoints for direct forecasting and scripts for fine-tuning on a team's own data.

SetupDocsCommunityMaturity

Verdict Our Kronos environment occupied 4,791 MB, and pytest passed 4 tests before a fifth file failed during collection because parametrize received a zip. Use it as a research candidate when candlestick tokenization is relevant and you can run strict out-of-sample, cost-aware comparisons.

Full review →

164

gs-quant

goldmansachs/gs-quant · ★ 13k ▲54/7d · Python

GS Quant is a Python library for derivative analytics, market data, risk work, and quantitative trading research. Some statistical code can run locally, while access to Goldman Sachs APIs and Marquee data requires institutional client credentials.

SetupDocsCommunityMaturity

Verdict Our gs-quant install failed after 78 seconds because its configured Goldman Sachs package index could not be resolved, so the source checkout was not ready for a clean public-network environment. Use it if your firm already has Marquee credentials and wants Goldman Sachs analytics, data, and session APIs in Python.

Full review →

165

Crucix

calesthio/Crucix · ★ 12k ▲57/7d · JavaScript

Crucix is a self-hosted dashboard that polls 27 public-data and OSINT sources every 15 minutes, then combines news, markets, aircraft, fires, radiation, satellites, and conflict signals in one interface. Optional LLM, Telegram, Discord, and Alpaca connections add summaries, alerts, commands, and trading actions.

SetupDocsCommunityMaturity

Verdict Our Crucix install took 14 seconds and 31 MB, but there was no build or test target and npm audit found 6 known vulnerabilities, so this is a localhost experiment rather than an intelligence system to trust. Its 27-source dashboard is useful for spotting topics to investigate.

Full review →

166

SQLBot

dataease/SQLBot · ★ 6.8k ▲39/7d · JavaScript

SQLBot's primary README is Chinese, and a shorter English translation is available. It is a self-hosted ChatBI application that turns plain-language questions into SQL, tables, charts, and follow-up analysis against connected business data. RAG, terminology, prompts, and saved SQL examples help it understand a team's database vocabulary.

SetupDocsCommunityMaturity

Verdict Our SQLBot run installed 241 packages and consumed 5,405 MB, then pytest discovered 0 tests and pip-audit reported 30 known vulnerabilities, so adoption needs a real security and regression budget. Use it when you want a complete, actively maintained ChatBI workspace and can lock down its model, database, and container boundaries.

Full review →

167

quant-mind

LLMQuant/quant-mind · ★ 2.9k ▲53/7d · Python

QuantMind is a Python framework that turns financial papers and news into typed, timestamped knowledge with citations attached. It gives retrieval systems and coding agents more traceable inputs than a folder of loosely parsed documents.

SetupDocsCommunityMaturity

Verdict Our QuantMind run passed 485 tests, but its 893 MB environment contained 24 known vulnerabilities, so adoption should start with dependency triage rather than production data. Try it when cited paper or PR Newswire artifacts are the immediate requirement and an agent-friendly checkout helps your team.

Full review →

168

PLFM_RADAR

NawfalMotii79/PLFM_RADAR · ★ 25k ▲122/7d · PLSQL

AERIS-10 is an alpha-stage open hardware design for a 10.5 GHz pulsed linear-frequency-modulated phased-array radar, with FPGA and microcontroller firmware plus a Python display. It gives radar researchers schematics, board files, signal-processing logic, and software to study or extend instead of starting every RF and digital subsystem from scratch.

SetupDocsCommunityMaturity

Verdict Our PLFM_RADAR software check installed 34 packages and built in 17 seconds, but it had no test target and did not validate any radar hardware. Use AERIS-10 as a serious engineering reference or research collaboration, not as order-ready production files.

Full review →

169

TVAPP

youhunwl/TVAPP · ★ 23k ▲279/7d · JavaScript

TVAPP is a Chinese-language catalog of Android TV applications, APK downloads, interface sources, and live-TV sources, not a conventional JavaScript application you build and deploy. No English documentation is evident in the supplied README excerpt. It solves the practical problem of finding TV-focused media, karaoke, gaming, and utility apps in one maintained index.

SetupDocsCommunityMaturity

Verdict Our 2026-09-06 run could not clone TVAPP, so we never reached installation, build, or tests. Treat it as a fast-moving Chinese download directory, not as a verified software distribution or a reproducible open-source application.

Full review →

170

semantica

semantica-agi/semantica · ★ 13k ▲521/7d · Python

Semantica is a Python toolkit for turning documents and enterprise data into knowledge graphs, decision records, provenance trails, and policy-aware context for AI agents. It sits beside an LLM or agent framework so teams can trace the data, rules, relationships, and recorded outcomes around a decision without claiming access to the model's private reasoning.

SetupDocsCommunityMaturity

Verdict Our Semantica install used 6,801 MB and its test run ended with 159 failures plus 41 collection or setup errors, so this is a platform to evaluate module by module, not adopt wholesale. Its graph, provenance, decision, and standards support is unusually broad for teams that truly need those pieces together.

Full review →

171

TradingView-API

Mathieu2301/TradingView-API · ★ 5.0k · JavaScript

TradingView-API is an independent Node.js client for reading TradingView market prices, chart periods, indicators, drawings, and replay data. It gives JavaScript programs access to chart sessions and studies that developers otherwise use through TradingView's website.

SetupDocsCommunityMaturity

Verdict Our TradingView-API run passed 50 of 66 tests in 20 seconds, but npm audit found 17 vulnerabilities, including 1 critical and 12 high, so it belongs in an isolated research or alerting service before it belongs near money. Use it when TradingView-specific indicators or replay data justify watching an unofficial WebSocket client for breakage.

Full review →

172

tick-stock-panel

shy3130/tickflow-stock-panel · ★ 4.7k ▲216/7d · Python

TickFlow Stock Panel is a self-hosted Chinese-language workstation for screening, monitoring, and backtesting mainland China A-share stocks. Its primary README and linked operating docs are Chinese, with no English guide linked; the browser app combines local data, strategy rules, alerts, charts, and optional AI analysis.

SetupDocsCommunityMaturity

Verdict Our TickFlow Stock Panel run finished with 28 passes, 20 failures, and 69 collection or setup errors, so this is a codebase to repair before using for research decisions. Chinese-reading hobbyists may still value its broad A-share workflow and local data design.

Full review →

173

desktop-fly

DenisSergeevitch/desktop-fly · ★ 972 ▲188/7d · Swift

DesktopFly is a macOS desktop pet whose movement is driven partly by a spiking simulation built from FlyWire fruit-fly connectome data. It turns a neuroscience dataset into a visible toy: the fly walks on window edges, reacts to cursor movement, and exposes an interactive view of its modeled neural circuit.

SetupDocsCommunityMaturity

Verdict Our 4-second Python-side build succeeded, but no test target ran and the macOS Swift application was outside that sandbox result. DesktopFly is worth trying as a compact, inspectable science toy on macOS, especially because the README clearly separates real connectivity from modeled behavior.

Full review →

174

projeto-dados-ia-databricks

lvgalvao/projeto-dados-ia-databricks · ★ 277 ▲7/7d · Python

Projeto Dados IA Databricks is a Portuguese-language workshop for building a fictional perfume distributor's analytics stack on Databricks Free Edition; no English documentation is provided. It teaches SQL, medallion layers, Asset Bundles, dashboards, Genie, and Claude Code through seeded business data with deliberate quality problems.

SetupDocsCommunityMaturity

Verdict Our run installed 35 packages in 11 seconds and built in 3 seconds, but the repository supplied no automated test target. Portuguese-speaking beginners can learn a lot from its deliberate dirty data and exact Databricks failure notes.

Full review →

175

social-analyzer

qeeqbox/social-analyzer · ★ 24k ▲73/7d · JavaScript

Social Analyzer is an OSINT tool that checks a username or name across more than 1,000 social sites and returns possible profile matches. It ships as a Node web app and CLI plus Python entry points, with optional browser automation, screenshots, OCR, metadata extraction, and search-engine lookups.

SetupDocsCommunityMaturity

Verdict Our Social Analyzer run installed 415 packages but failed its only npm test in 7 seconds, and the audit found 26 known vulnerabilities, so this should be an analyst-controlled research aid rather than a trusted service. Its broad site list, JSON output, and several interfaces are useful for leads.

Full review →

176

WeMM-Embedding

Tencent/WeMM-Embedding · ★ 1.6k ▲171/7d · Python

WeMM-Embedding is Tencent's family of models for putting text, images, video, scanned documents, and mixed media into the same searchable vector space. It is aimed at retrieval systems that need one representation across media instead of separate text and vision indexes.

SetupDocsCommunityMaturity

Verdict Our WeMM-Embedding install pulled 92 packages, occupied 5,047 MB, and exposed 3 known vulnerabilities before any model weights were downloaded. Trial it when one index must retrieve across text, images, video, and visual documents, and compare its vectors on your own queries.

Full review →

177

minio

minio/minio · ★ 61k · Go

MinIO is an S3-compatible object-storage server for keeping files and large data sets on infrastructure you control. This open-source repository is now archived and unmaintained; MinIO directs new users to separate AIStor Free and Enterprise products.

SetupDocsCommunityMaturity

Verdict Our MinIO build succeeded, but 45 of 46 Go packages passed before the test run failed, and the repository is now archived. Do not choose this frozen upstream for a new production object store.

Full review →

178

TelegramGroup

AZeC4/TelegramGroup · ★ 23k ▲62/7d

TelegramGroup is a Simplified Chinese directory of Telegram groups, channels, bots, account guides, and paid promotions; it has no English documentation. It is a very large README, not software you install, and its job is to help Chinese-speaking users browse Telegram links by topic.

SetupDocsCommunityMaturity

Verdict Our sandbox did not run TelegramGroup because it has no supported code ecosystem or Dockerfile, so judge it as a 250 KB link directory rather than an open-source application. Chinese-speaking users may find obscure groups and bots here, but the mixture of paid placements, referrals, financial offers, and sensitive categories demands manual checking before every click.

Full review →

179

lx-music-source

pdone/lx-music-source · ★ 8.8k · JavaScript

lx-music-source is documented in Chinese, and the repository has no English README. It is a directory of 8 third-party JavaScript source adapters that users can import into LX Music to resolve playable music URLs through outside services.

SetupDocsCommunityMaturity

Verdict Our lx-music-source run built in 7 seconds and pip-audit found 0 known vulnerabilities, but no test target exercised any of its 8 executable adapters. Treat it as an unverified link shelf for experienced LX Music users, not as a maintained playback service.

Full review →

180

getcontact-cli

xdreizein666/getcontact-cli · ★ 585 ▲39/7d · Python

GetContact CLI is a Python command-line client whose documentation is in Indonesian; the repository has no English guide. It looks up phone profiles and tags through GetContact's private API, checks account quota, handles captchas, and can process a CSV without the Android app or a browser.

SetupDocsCommunityMaturity

Verdict Our GetContact CLI install finished in 24 seconds with 35 packages and 37 MB on disk, but the repo had no test target, so the quick setup does not prove its private API still works. Use it only for authorized research with an account you can afford to lose and a plan for deleting the saved results.

Full review →

181

DeepSeek-V4-J-Space-Capability-Realization-Report

Tiger3807861189/DeepSeek-V4-J-Space-Capability-Realization-Report · ★ 1.0k

This benchmark report is primarily written in Chinese, and a full English translation is included. It claims to compare DeepSeek V4-Flash-Vision-Exp with and without J-Space V3.7 across selected agent and reasoning benchmarks, but the repository publishes summary tables rather than the harness, task traces, or result files needed to audit those claims.

SetupDocsCommunityMaturity

Verdict The v2.0 repository ships 2 report files and no executable harness, raw task traces, or machine-readable results, so its score table is not decision-grade evidence. Read it as a set of claims and experiment ideas, then reproduce the relevant subset before changing models or adopting J-Space.

Full review →