mrkeyoor.com_
Tue 01 Sept 17:44 UTC
AI Toolsevaluationupdated 23 Aug 2026

buildwithclaude review

Build with Claude is a directory and installable marketplace for Claude Code agents, commands, hooks, skills, plugins, and links to MCP servers. It helps developers find reusable Claude Code extensions without hunting through scattered repositories, while its web interface provides search and filtering.

+36stars / 7d
Verdict

Build with Claude is useful as a search layer and a starting point for teams already committed to Claude Code. Do not mistake catalog size for curation depth: install selected entries after reading their permissions and code, especially hooks. The 879 MB install and 42 audit findings make the repository itself a poor choice for teams that need a lean, clean dependency baseline.

We ran it

Lab card: what happened when we ran buildwithclaude
Install✓ · 34s925 packages · 879 MB
Buildn/ano build script
Tests✓ · 15s3 passed · 0 failed of 3 (node:test)
Known vulns421 critical · 19 high · 20 moderate · 2 low (npm audit)
Repo2173 files~73,385 lines of source · 23 MB · 2 CI workflows · Dockerfile · tests dir

Answers from our run

Does buildwithclaude build from source?

Dependencies installed in 34 seconds (925 packages), and the project has no separate build step. We cloned commit d16bece into a clean Debian container with 3 CPUs and no project-specific setup.

Do buildwithclaude's tests pass?

Yes: 3 of 3 passed when we ran the project's own test command (node:test). Some failures need services or credentials a bare container does not have.

Does buildwithclaude have known vulnerabilities in its dependencies?

npm audit flagged 42 known advisories in the dependency tree, including 1 critical at the time of our run.

Who should not use buildwithclaude?

Developers who want a small dependency footprint: our checkout used 879 MB after installing 925 packages.

What are the alternatives to buildwithclaude?

Awesome Claude Code, Awesome Claude Code Subagents, Claude Code. Build with Claude is useful as a search layer and a starting point for teams already committed to Claude Code.

Setup3/5Easy marketplace command, heavy local install and plugin-specific needs
Docs4/5Clear install, usage, plugin formats, and contribution instructions
Community4/5Pushed on 2026-08-23 with an active submission queue
Maturity3/5Working catalog and tests, but audit debt and uneven release history

Who it’s for

Claude Code users who want a browsable catalog plus copyable plugin install commands.
Teams standardizing a shared set of agents, commands, skills, or hooks.
Plugin authors who want a documented contribution format and marketplace listing.
Developers willing to inspect an extension before giving it tools or hook access.

Who it’s NOT for

Developers who want a small dependency footprint: our checkout used 879 MB after installing 925 packages.
Security-sensitive teams expecting a clean npm audit: our run found 42 known vulnerabilities, including 1 critical and 19 high.
Anyone who treats a marketplace listing as a security review: the repository accepts community submissions, and plugins can declare tools or run hooks on Claude Code events.
Users outside Claude Code who expect a general agent package manager: the install commands and contribution formats are built around Claude Code.
Operators expecting a versioned release trail for the whole catalog: the latest GitHub release is a 2025 CLI package tag, while repository work has continued since then.

Setup reality

Our fresh Node 22 container installed 925 packages in 34 seconds and occupied 879 MB. The repository had no build script, so there was no build step to run. Its test command finished in 15 seconds with 3 node:test cases passing and none failing. npm audit reported 42 known vulnerabilities: 1 critical, 19 high, 20 moderate, and 2 low.

Using the marketplace itself requires Claude Code, then /plugin marketplace add davepoon/buildwithclaude. Individual plugins may need their own accounts, tokens, local binaries, or services, so the three-line quick start is only the catalog setup. Running the web interface locally uses the Node workspace and its start script.

Hooks deserve special care because they run on lifecycle events, and agents or commands can receive tool access declared in their files. Review each selected plugin rather than installing the all-in bundles by default. The large dependency tree and audit result also make a pinned, reviewed deployment preferable to an unattended install.

A catalog for Claude Code extensions

Claude Code has several extension types, and finding them is messy. Agents, slash commands, hooks, skills, full plugins, and MCP integrations often live in separate repositories with different installation notes. Build with Claude gathers those pieces into one marketplace and backs it with a searchable website. You can browse by type, inspect documentation, copy an install command, or add the whole repository as a Claude Code marketplace.

The repository also maintains its own bundled collections. The README lists 117 agents, 175 commands, 28 hooks, 26 skills, and 51 plugin packages. Its web index reaches beyond those bundled files to community plugins, MCP servers, and other marketplaces. Those catalog counts come from the project documentation; they are discovery inventory, not proof that every entry has received the same review.

That distinction matters. Build with Claude is most useful as a map. It cuts down search time and gives contributors a consistent place to submit work. It does not remove the need to inspect what an extension can read, write, execute, or send elsewhere.

Installing from the marketplace

The normal path is short. Add davepoon/buildwithclaude with Claude Code's marketplace command, search the marketplace, then install a named plugin. Bundles exist for all agents, all commands, and all hooks. Manual installation is also documented: clone the repository and copy selected Markdown definitions into the matching Claude directory.

Selective installation is the sensible default. An agent file can declare access to tools such as file writing or Bash. A hook may run before or after tool calls, when a session starts, or when it ends. Some entries are simple instruction files, while others connect to external services or ship executable code. The marketplace format records metadata and source locations, but that record is not a permission prompt or an independent audit.

Teams should treat each plugin like a development dependency. Read its source, confirm its tool list, check any network destinations, and pin the version you approved. The all-in bundles are convenient for exploration, but they widen the review surface immediately.

What happened when we ran it

We cloned commit d16bece into a fresh Debian container with Node 22, 3 CPUs, 8 GB of RAM, no secrets, and an unprivileged user. The checkout contained 2,173 files, about 73,385 lines of source, and used 23 MB before installation.

npm installed 925 packages in 34 seconds. The installed workspace occupied 879 MB, a surprisingly large result for a repository whose visible product is mostly a catalog and web interface. There was no build script or target, so we skipped the build rather than inventing one. The available node:test run completed in 15 seconds: 3 tests passed and none failed.

The uncomfortable result was npm audit. It reported 42 known vulnerabilities: 1 critical, 19 high, 20 moderate, and 2 low. That does not prove an exploitable path in every use of the catalog, but it is too much debt to wave through. If you run the local web interface or automation scripts, review the dependency paths and decide which findings touch your deployment before exposing it.

The contribution model is clear

The contributor guide is more specific than the front page. Agents have required names, descriptions, categories, and a prescribed opening. Commands document arguments and allowed tools. Hooks name their lifecycle event, matcher, implementation language, and requirements. Contributors are told to check for overlap, keep one purpose per component, run validation, and submit a pull request.

That structure helps maintain a repository with many small text-based extensions. The root package includes validation scripts, registry generation, plugin fetching, unit tests, and an integration harness. There are also two CI workflow files, a Dockerfile, a tests directory, and an npm workspace for the web interface. The project is doing more than collecting links in a README.

Still, the public submission queue shows the editorial challenge. Open items include requests to add MCP servers, community plugins, and adjacent tools. A maintainer can verify shape and placement more easily than it can verify every prompt's judgment or every hook's behavior. Competent developers should expect uneven quality across a collection this broad.

Health and release signals

The repository was pushed on 2026-08-23, the day of our review, and its open count was 12 issues and pull requests. Recent issue updates include new MCP server and plugin submissions, so the queue reflects ongoing catalog work rather than a silent project. The MIT license is straightforward for the repository's own code and content, though an indexed external plugin can have its own license and terms.

The latest GitHub release tells a less tidy story. It is bwc-cli@1.2.4, published on 2025-10-24, rather than a current release for the whole marketplace. The root package also reports version 1.0.0 while the marketplace metadata has its own version. Current pushes show active work, so the old release tag is not evidence of abandonment. It does mean teams should pin a commit or plugin version instead of assuming the GitHub releases page describes the present catalog.

Who should use it

Use Build with Claude when your team already uses Claude Code and wants one place to compare extensions or publish an internal selection. Its install path is easy, its contributor documentation is concrete, and the website makes a large ecosystem easier to search.

Walk away if your policy requires every dependency and automation hook to arrive pre-vetted. The project cannot make that promise, and our audit result raises a separate concern in its own Node workspace. For most teams, the right approach is to use the site as a directory, install only a few reviewed plugins, and keep an approved list in your own repository.

Alternatives

ProjectWhat it isPick it when
Awesome Claude Code gh↗A curated list of Claude Code commands, workflows, tools, and learning resources.pick this instead when you want a reading list and examples rather than an installable marketplace.
Awesome Claude Code SubagentsA focused collection of specialist Claude Code subagent definitions.pick this instead when subagents are your only target and a narrower catalog is easier to audit.
Claude Code gh↗The official Claude Code repository, documentation, issue tracker, and plugin host application.pick this instead when you need the official client and supported plugin mechanics rather than community discovery.

What people are saying

  1. [github-trending] davepoon/buildwithclaude

Sources

  1. Build with Claude README
  2. Build with Claude repository facts
  3. Build with Claude contribution guide
  4. Build with Claude package scripts
  5. Build with Claude latest release
  6. Build with Claude issue tracker

More ai tools reviews

claudian · SkillSpector · robin · mjlab · MoGe · awesome-design-md · the whole board →