Search API

Combined Knowledge Search

The recommended search endpoint searches across all data types in a single request.

POST /api/v1/search/knowledge

{
    "query": "search term",
    "workspace_id": "optional",
    "document_tags": ["optional-filter"],
    "page": 1,
    "per_page": 10
}

Response:

{
    "success": true,
    "results": {
        "notes": [...],
        "memories": [...]
    },
    "total_found": {
        "notes": 5,
        "memories": 3
    },
    "page": 1
}

This endpoint queries both notes and memories in parallel using Typesense BM25 search.

• workspace_id is optional — omit to search notes across all workspaces the user can access
• Use document_tags to filter results by tags

Type-Specific Search

For searching within a single type:

Type	Endpoint
Notes	POST /api/v1/files/notes/search
Memories	POST /api/v1/memories/search
URLs	via search_urls MCP tool

AI Chat

POST /api/v1/chat

{
    "query": "What did I save about Redis caching?",
    "conversation_id": "optional-for-context",
    "project_id": "optional-project-filter"
}

Uses semantic search across all collections, builds context from top results, and generates a response via the configured LLM provider (OpenAI, Google Gemini, Groq, or Cerebras).