context7-skill plugin
Library docs from Context7 over REST: about a hundred tokens on demand and no always-on MCP server schemas in context
- 91 ★ on the repository
- Last commit
Extended access
Why this level
- The skill calls the external Context7 REST API and sends it library names and query topics.
Install
- Works in
Manual install
/plugin marketplace add netresearch/claude-code-marketplaceThe first README option: add the Netresearch marketplace, then `/plugin install context7@netresearch-claude-code-marketplace`.
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- The skill calls the external Context7 REST API and sends it library names and query topics.
This is third-party code. Review the repository files before installing.
What it does
The Netresearch skill fetches up-to-date library documentation from Context7 through its REST API and needs no MCP server. MCP tool schemas permanently occupy context, while the skill adds around a hundred tokens and only when a question concerns a library. It triggers on phrases like "how do I use React hooks", "Next.js App Router API" or "Prisma query patterns". The work is done by the `scripts/context7.sh` wrapper: `search "prisma"` finds a library ID, and `docs "/prisma/prisma" "queries"` returns documentation on a topic such as hooks, routing or middleware. The README claims more than fifty libraries, including React, Next.js, Vue, Express and Prisma. The `CONTEXT7_API_KEY` key is optional and gives a higher request limit.
Who it is for. Developers who occasionally ask an agent about library APIs and do not want a full Context7 MCP server in the context.
Good fit when
- The agent must answer from the current Next.js App Router API, and stale model memory falls short
- You need Prisma query patterns found in fresh documentation on the topic
- The agent context is tight and always-on MCP tool schemas cost too much
- Library documentation has to be checked by hand with `scripts/context7.sh search`
Not a fit when
- You need docs constantly and in volume, where a full Context7 MCP server is more convenient
- Queries cannot be sent to an external service, as the skill calls the Context7 REST API
- The library is not in the Context7 database, and the search will return nothing
Example request
How do I use React hooks? Fetch current documentation through context7 on the hooks topicLimitations
The skill works only with libraries present in the Context7 database and needs access to its REST API. Without a `CONTEXT7_API_KEY` key a lower request limit applies. Library queries go to an external service. The skill code is MIT-licensed and the content is CC-BY-SA-4.0, so keep that in mind when copying. Updating a no-marketplace install is manual with `git pull`, and `claude plugin update` does not work for that route.
How to disable. Run `/plugin uninstall` for the plugin or delete the `~/.claude/skills/context7` directory. Remove the `CONTEXT7_API_KEY` variable from the environment.
FAQ
How do I install the skill in Claude Code?
Add the marketplace with `/plugin marketplace add netresearch/claude-code-marketplace` and run `/plugin install context7@netresearch-claude-code-marketplace`.
Is an API key required?
No, `CONTEXT7_API_KEY` is optional. A key gives a higher request limit.
What is the advantage over the Context7 MCP server?
It keeps no tool schemas in context, using about a hundred tokens on demand against five hundred to two thousand always. It fits when docs are needed occasionally.
How do I call the script by hand?
Run `scripts/context7.sh search "prisma"` to find an ID and `scripts/context7.sh docs "/prisma/prisma" "queries"` for topic documentation.
How do I add the skill to a PHP or Node project?
In PHP use `composer require netresearch/context7-skill` with the netresearch/composer-agent-skill-plugin plugin. In Node you need `@netresearch/agent-skill-coordinator` and the package from GitHub.
Discussion
No questions or reviews yet.
Related
PluginsUp-to-date, version-specific library docs and code examples in your agent's context, via MCP or a CLI plus skill
Local documentation for agents: an MCP server searches a library of doc packages for 100+ libraries and answers offline
Context Engine: code search and memory for an agent
Context Engine
A skill that teaches an agent to query the Context Engine code index and symbol graph before reading files directly
An MCP server that turns a large codebase into a semantic feature graph for precise navigation and meaning-based search
A context profiler: a proxy captures an agent's requests to the model and shows what each request is made of and where tokens go.
The chub CLI gives an agent curated, versioned API docs by language, keeps local notes and takes ratings; a skill and an MCP server are included
README in short
The README presents the skill as a lightweight replacement for the Context7 MCP server, listing features, six install routes, sample prompts, manual script use and a context cost comparison table.