LLM integration

LogTape ships with built-in support for AI coding assistants through the Agent Skills standard. This means tools like Claude Code, GitHub Copilot, Cursor, Windsurf, and many others can automatically learn how to use LogTape correctly when working on your project.

What the skill provides

The bundled agent skill teaches AI assistants:

• How to get loggers with hierarchical categories
• The structured message syntax (named placeholders, not string interpolation)
• Configuration rules (configure() and configureSync())
• The library author rule (never call configure() in library code)
• Context management with with(), withContext(), and lazy()
• Sink and formatter setup
• Adaptor integration for existing loggers (winston, Pino, log4js)
• Testing patterns
• Common mistakes to avoid

This complements the llms.txt file, which provides a broad overview of the library for general-purpose LLM consumption.

Setting up with skills-npm

skills-npm by Anthony Fu scans node_modules for packages that declare skills and symlinks them into the appropriate agent directories.

1. Install skills-npm as a dev dependency:

   npm

   npm install -D skills-npm

   pnpm

   pnpm add -D skills-npm

   Yarn

   yarn add -D skills-npm

   Bun

   bun add -D skills-npm

2. Add a prepare script to your package.json:

   {
     "scripts": {
       "prepare": "skills-npm"
     }
   }

3. Run npm install (or your package manager's equivalent). This symlinks skills from node_modules into the appropriate agent directories (e.g., .claude/skills/, .cursor/skills/).

4. Add the generated symlinks to your .gitignore:

   skills/npm-*

After this one-time setup, the skill stays in sync with the installed version of LogTape. Every npm install refreshes the symlinks automatically.

Setting up with npm-agentskills

npm-agentskills by onmax is another tool that discovers agent skills from npm packages. It offers a CLI and a Nuxt module.

CLI usage

npx agents export --target claude

You can also export to other agents:

npx agents export --target cursor
npx agents export --target codex

To see all discovered skills:

npx agents list

Automatic setup via postinstall

Add to your package.json:

{
  "scripts": {
    "postinstall": "agents export --target claude"
  }
}

Nuxt integration

For Nuxt projects, add the module and skills are discovered automatically when running nuxi prepare or nuxi dev.

Manual setup

If you prefer not to use a discovery tool, you can set up the skill manually for your specific AI coding assistant.

Claude Code

Copy or symlink the skill directory into your project:

mkdir -p .claude/skills
cp -r node_modules/@logtape/logtape/skills/logtape .claude/skills/logtape

Cursor

mkdir -p .cursor/skills
cp -r node_modules/@logtape/logtape/skills/logtape .cursor/skills/logtape

Other agents

Most Agent Skills-compatible tools look for skills in a directory like .<agent>/skills/ in your project root. Consult your tool's documentation for the exact path.

Setting up for Deno

For Deno users who install LogTape from JSR, the skill file is accessible directly from the repository. You can reference it by URL or copy it into your project's skill directory manually.

How it works

The @logtape/logtape package.json declares the skill using the agents field:

{
  "agents": {
    "skills": [
      { "name": "logtape", "path": "./skills/logtape" }
    ]
  }
}

When a discovery tool like skills-npm or npm-agentskills runs, it reads this field from every package in node_modules and copies or symlinks the skill directories so that agent tools can discover and load them. The skill's description field in its frontmatter tells the agent when to activate it, so it loads automatically whenever the agent is working on logging-related code.

NOTE

The agents field is purely metadata. It does not add any runtime dependencies or affect how LogTape works.

Also available: llms.txt

LogTape's documentation site also generates an llms.txt file following the llms.txt standard. This file provides a comprehensive overview of the entire library and is useful for general-purpose LLM tools that support the standard.

While the agent skill focuses on how to write code with LogTape (practical rules and common mistakes), llms.txt focuses on what LogTape is (API surface, concepts, and documentation links).