Guide · Agent skill
How to use the Documentation Lookup agent skill (Context7)
Updated · written with AI from the project's own documentation and checked against our catalogue data
Documentation Lookup is an agent skill that tells your AI agent to check live library documentation before answering questions about a framework, library, or API. It relies on Context7, a documentation service connected to the agent through MCP (a standard way to plug external tools into an agent). This guide explains what the skill does, how to install it, how to trigger it, and what to watch for.
Documentation Lookupaffaan-m/ECC · MIT Download the skill Quick start
- 1Get the ZIP. Download the documentation-lookup ZIP from this skill's page. It contains SKILL.md and agents/openai.yaml.
- 2Install the skill. Unzip it into ~/.claude/skills/ for all projects, or into .claude/skills/ inside a project. In the Claude apps, upload it under Settings → Capabilities → Skills.
- 3Configure Context7 MCP. Make sure the Context7 MCP server is set up in your agent environment, such as Claude Code, Cursor, or Codex, and that the machine has network access.
- 4Ask a library question. Ask a setup, API, or code question that names the framework or library, such as Next.js, Prisma, or Supabase.
- 5Check the answer. Look for the library and version the agent cites, and review any code examples taken from the docs before you use them.
What the skill makes the agent do
When you ask about a library, framework, or API, the skill directs the agent to look up the matching library in Context7 first. The agent then reads the current documentation and answers from it, instead of from what it remembers from training, which may be out of date.
The skill uses two Context7 tools. resolve-library-id finds the Context7 library ID for the library you named, such as /vercel/next.js. query-docs then fetches documentation and code snippets for that ID and your question.
- Setup and configuration questions about fast-changing frameworks
- API references and method lists
- Code that depends on a specific library
Install the skill (ZIP download or Claude apps)
Download the ZIP from the skill's page on this site. The skill package contains two files: SKILL.md and agents/openai.yaml.
For Claude Code, unzip the package into your personal skills folder at ~/.claude/skills/, or into a project's .claude/skills/ folder to make it available only in that project.
In the Claude apps, you can upload the ZIP instead under Settings → Capabilities → Skills.
- Personal install: ~/.claude/skills/
- Project install: .claude/skills/ inside your project
- Claude apps: Settings → Capabilities → Skills
Set up the Context7 MCP server first
The skill cannot fetch documentation without the Context7 MCP server. Configure it in your agent environment before using the skill. The skill's source lists Claude Code, Cursor, and Codex as harnesses where the Context7 MCP can be configured.
The skill also needs network access to Context7 to query documentation. Without the server, the skill will not return results.
- Context7 MCP server configured in the agent environment
- Network access to Context7
Trigger the skill with natural requests
You do not need a special command. The skill activates when your request involves setup questions, API references, code examples, or a named framework. Naming the library in your question helps the agent find the right match.
Requests that fit the skill include setup questions, code that depends on a library, API or reference questions, and any mention of a specific framework such as React, Vue, Svelte, Express, Tailwind, Prisma, or Supabase.
- How do I configure Next.js middleware?
- Write a Prisma query that loads a user with their posts.
- What are the Supabase auth methods?
Examples of how the agent uses it
For a Next.js question, the agent calls resolve-library-id with the library name Next.js and your question, picks the best match (for example /vercel/next.js), then calls query-docs with that ID and your question. It answers using the returned snippets and can include a minimal middleware.ts example from the docs.
For Prisma, the agent resolves the official Prisma library, fetches docs on your query, and returns the matching Prisma Client pattern, such as include or select, with a short snippet.
- Next.js: resolve the library, query middleware setup, answer with a middleware.ts example
- Prisma: resolve the official library, query relations, return include or select usage
- Supabase: resolve the library, query auth methods, summarize with examples
Safety notes: network use and secrets
This skill does not run scripts, but it does use the network. The agent sends your question to Context7, an external service, so anything in the question leaves your machine.
Redact API keys, passwords, tokens, and other secrets from your questions before asking. The skill's own guidance says to treat the question as potentially containing secrets before it is passed to resolve-library-id or query-docs.
- Runs scripts: no
- Uses network: yes
- Remove secrets from questions before sending them
Tips for better results
Use your full question, since it is used to rank results. When you mention a version, such as React 19 or Next.js 15, the agent prefers a version-specific library ID if one is listed. The skill limits lookups to three calls per question; if the answer is still unclear, the agent should say so rather than guess.
For general programming questions with no named library, expect the lookup to add delay without much benefit.
- Name the library and version in your question
- Ask a specific question rather than a broad one
- Expect at most three lookups per question
What people use it for
- Checking the current setup steps for a framework before configuring it
- Writing code that calls a specific library and needs accurate method names
- Looking up the auth methods or configuration options of a backend service
- Finding the right query pattern in an ORM such as Prisma
Frequently asked questions
Do I need to install anything else to use this skill?
Yes. The skill needs the Context7 MCP server configured in your agent environment, plus network access to Context7. The skill cannot fetch documentation without it.
Is the documentation lookup skill free?
The skill is published under the MIT licence, as stated in its SKILL.md. The page does not state pricing for Context7, so check Context7 directly for its terms.
Which agents does it work with?
The skill's source says it applies across harnesses that have the Context7 MCP configured, and names Claude Code, Cursor, and Codex as examples.
Does the skill run scripts on my computer?
No. The listing says it runs no scripts, but it does use the network to send your questions to Context7.
Why does the agent say it is unsure after a lookup?
The skill allows at most three resolve or query calls per question. If the answer is still unclear after that, the skill tells the agent to state the uncertainty rather than guess.
When should I not use this skill?
Skip it for general programming questions that do not name a library, since the lookup adds delay without benefit. Also avoid sending questions that contain secrets.
New agent skills, repos and AI tools — every week
The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.
Our verdict
Use this skill if you work with fast-changing frameworks or need exact API details for a specific library, and you already have the Context7 MCP server set up. Skip it if you do not have Context7 configured, if your questions do not involve a named library, or if you cannot keep secrets out of your prompts.
Download the skill