Skip to main content
Use search_tools to query the UseAgents registry with a plain-language request. This helps your agent discover real tools instead of fabricating package names or relying on stale memory.
Requires an API key. Configure the x-api-key header when connecting the MCP server.

​
Parameters

ParameterTypeRequiredDescription
qstringYesNatural-language query. Searches across tool names, descriptions, tags, languages, and categories.

​
Query guidance

Good queries usually include:
  • The problem to solve
  • The language or framework
  • Important constraints such as deployment model or integration surface
Example queries:
  • authentication library for Node.js
  • python sdk for Slack bots
  • self-hosted feature flag service for Go

​
Returns

The tool returns results in Toon (Token-Oriented Object Notation), a compact, token-efficient format for LLM consumption. Learn more at toonformat.dev. Each result includes:
FieldTypeDescription
namestringTool name
descriptionstringFull description
languagesstring[]Supported languages
categoriesstring[]Tool categories
slugstringURL-friendly identifier for the tool

​
Example call

{
  "q": "authentication library for Node.js"
}
Example response (Toon format): 
results[1]:
  - name: Example Auth SDK
    description: A TypeScript authentication library for Node.js applications.
    languages[2]: TypeScript,JavaScript
    categories[2]: Authentication,SDK
    slug: example-auth-sdk

​
Use this tool when

  • The agent needs to recommend a library, CLI, SDK, or API
  • The agent should avoid fabricating or guessing tool names
  • The agent needs accurate metadata before selecting a dependency

​
Typical workflow

  1. Call search_tools with a task-focused query.
  2. Review the ranked results and metadata.
  3. Use the returned links and descriptions to narrow the candidate set.
  4. Only then recommend or integrate the selected tool.