Anthropic Agent Skills
anthropics/skillsAnthropic's reference skills repo: 20 skills including document tools (docx, pptx, xlsx, pdf), frontend-design, skill-creator and mcp-builder.
Python tool and MCP server that converts documentation sites, repos, PDFs and other sources into agent skills and RAG-ready knowledge, with 22 export targets.
A useful, tested conversion tool, but the real work is in a Python package and MCP server you must install, and the repo's skill count is inflated by test fixtures.
Skill Seekers is a conversion pipeline more than a skill. You point it at a documentation site, a GitHub repository, a PDF, a video or one of about 18 source types, and it builds structured knowledge assets and packages them as a Claude skill, or for other targets such as Gemini, OpenAI, LangChain, LlamaIndex or Cursor rules (the README counts 22 targets). It detects conflicts between sources, for example docs that disagree with the code. The agent-facing skill, skill-builder, is a small 4 KB file that says to use the Skill Seekers MCP server, which exposes 40 tools, and falls back to the command line if the MCP server is missing. An Anthropic API key can be set to let the tool enhance output, and without one you can package the result and upload the zip to claude.ai manually. The repo is large (about 3,100 files, mostly code and golden test fixtures), has CI and a test suite, and 26 SKILL.md files, many of which are test fixtures rather than shippable skills. Installation involves a Python package, not just dropping in a SKILL.md.
A thin 4 KB agent skill, but a large Python codebase with about 44 reference files, golden test fixtures and CI behind it.
Counted from the repository file tree. How we score depth.
The reviewable files are two short SKILL.md stubs and test fixtures that tell the agent to use the Skill Seekers MCP tools or CLI to scrape sources and package skills. All behavior lives in a separate Python package installed with an unpinned pip command, so what it fetches, uploads or writes could not be checked here.
The repository (3138 files) on Oct 29, 2026. Not a full review: 2998 of 3138 files were not downloaded (large repository; we reviewed the skill folders); it relies on code outside the repository: The skill-seekers PyPI package and its MCP server (40 tools), which perform the actual scraping, enhancement and upload; About 2,998 readable repository files (Python source, workflows, plugin manifests, examples) that are not in the local copy; Remote config sources used by fetch_config, and the sites, repos and videos the user asks it to scrape.
NVIDIA SkillSpector static scan: 0 pattern hits. SkillSpector reported no hits on the files we have (two skill stubs, the README and golden test fixtures).
A review, not a certification. Skills change after we check them — read the files before installing.
MIT — commercial use allowed. Check the license of any documentation you convert.
Yes. Scrape and build runs are command-line jobs that can be scripted, though they need a person to review the generated skill.
Anthropic's reference skills repo: 20 skills including document tools (docx, pptx, xlsx, pdf), frontend-design, skill-creator and mcp-builder.
Vercel Labs' open-source CLI (npx skills) for installing, updating and removing agent skills across Claude Code, Codex, Cursor, OpenCode and dozens more agents.
A Claude Code meta-skill that designs a project-specific agent team, writes the agent definitions and skills for it, and has an evolve skill to refine them from feedback.
A meta-skill that watches your sessions, logs corrections and repeated work, and proposes improvements to your other skills for you to approve.
Skill Seekers converts documentation websites, GitHub repositories, PDFs and other sources into agent skills and RAG data, with conflict detection between sources.
Install the Python package, then run the CLI against a docs URL or repo, or use its MCP server through the skill-builder skill. Without an API key you can package the output and upload the zip to claude.ai yourself.
No for basic use. An ANTHROPIC_API_KEY lets it enhance and upload skills automatically; otherwise you package the zip and upload it manually.
anthropics/skills includes a skill-creator for writing skills with the model. Skill Seekers instead automates scraping existing documentation and code into skills.