search_tools

When to use

Call search_tools whenever an agent needs to find the right developer tool for a task. The pipeline may return results immediately or ask clarification questions first — depending on how specific the query is and whether context filters are provided.

Input schema

query················································································stringrequired

Natural language search query describing the tool you need.

context················································································objectoptional

Optional pre-filters to skip clarification questions and jump directly to results.

context.language················································································stringoptional

Programming language filter (e.g. "TypeScript", "Python", "Go").

context.category················································································stringoptional

Tool category filter (e.g. "database", "testing", "monitoring").

context.license················································································stringoptional

License type filter (e.g. "MIT", "Apache-2.0", "proprietary").

context.deployment················································································stringoptional

Deployment model filter (e.g. "self-hosted", "cloud", "serverless").

Examples

Basic search

{
  "query": "vector database for embeddings"
}

I need a vector database for storing embeddings.
// Claude will call search_tools with:
// { "query": "vector database for embeddings" }

// In Cursor chat:
// "Find me a vector database for embeddings"
// Cursor calls search_tools automatically

{
  "query": "ORM for PostgreSQL",
  "context": {
    "language": "TypeScript"
  }
}

I need a TypeScript ORM for PostgreSQL.
// Claude will call search_tools with:
// { "query": "ORM for PostgreSQL", "context": { "language": "TypeScript" } }

// In Cursor chat:
// "I need a TypeScript ORM for PostgreSQL"
// Cursor calls search_tools with language context

{
  "query": "monitoring and observability",
  "context": {
    "language": "Go",
    "category": "monitoring",
    "license": "Apache-2.0",
    "deployment": "self-hosted"
  }
}

Response format

The response type field determines the shape of the payload. There are two possible response types:

Clarification needed

Returned when the query is ambiguous and the pipeline needs more information. Pass the query_id and your answers to search_tools_respond to complete the search.

{
  "type": "clarification",
  "query_id": "q_abc123def456",
  "questions": [
    {
      "key": "language",
      "question": "What programming language are you using?",
      "options": ["Python", "TypeScript", "Go", "Java", "Rust"]
    },
    {
      "key": "deployment",
      "question": "What deployment model do you prefer?",
      "options": ["cloud", "self-hosted", "serverless"]
    }
  ]
}

{
  "type": "result",
  "recommendation": {
    "name": "Prisma",
    "description": "Next-generation TypeScript ORM for PostgreSQL, MySQL, and SQLite",
    "category": "database",
    "language": "TypeScript",
    "license": "Apache-2.0",
    "health_tier": "healthy",
    "confidence": 0.94,
    "reasons": [
      "Strong TypeScript support with generated types",
      "Native PostgreSQL support with migrations",
      "Active community with 35k+ GitHub stars"
    ]
  },
  "alternatives": [
    {
      "name": "Drizzle ORM",
      "confidence": 0.87,
      "description": "Lightweight TypeScript ORM with SQL-like syntax"
    },
    {
      "name": "TypeORM",
      "confidence": 0.72,
      "description": "Mature ORM for TypeScript and JavaScript"
    }
  ]
}

Error codes

400················································································Bad Requestoptional

Invalid or empty query string. Ensure the query parameter is a non-empty string.

500················································································Internal Erroroptional

Pipeline processing error. An unexpected failure occurred in the search pipeline. Retry or contact support.

Related tools

• search_tools_respond — Answer clarification questions to complete a search
• get_stack — Build a compatible tool stack for a use case
• report_outcome — Report usage feedback to improve future results