docs_search is a built-in tool that makes Qorven’s documentation queryable by Qors themselves. Prime + any specialist with docs in its tool allowlist can call it.

​
Invocation

{
  "tool": "docs_search",
  "args": {
    "query": "how to rotate the encryption key",
    "max_results": 3
  }
}

​
Response

{
  "results": [
    {
      "url": "https://docs.qorven.ai/security/per-install-encryption",
      "title": "Per-install encryption",
      "snippet": "The encryption key is generated at install time and protects every encrypted secret...",
      "relevance": 0.91
    },
    {
      "url": "https://docs.qorven.ai/ops/encryption-key",
      "title": "Encryption key operations",
      "snippet": "...",
      "relevance": 0.78
    },
    {
      "url": "https://docs.qorven.ai/cli/admin",
      "title": "qorven admin",
      "snippet": "...",
      "relevance": 0.65
    }
  ]
}

​
How to use it in your Qor

In the system prompt: 
When a user asks about Qorven features, install, or config:
1. Call docs_search with their question
2. Read the top 1-2 results
3. Answer using those as the source of truth
4. Cite URLs so the user can follow up

​
Hybrid ranking

Same pgvector + BM25 RRF stack used for memory search. Tuned on a fixed corpus (the docs), re-indexed on each docs deploy.

​
Alternative: fetch yourself

If your Qor has web_fetch but not docs_search, it can fetch https://docs.qorven.ai/llms.txt then pick + fetch individual pages. Slower, more tokens, but works without the tool.

Agent guide

All the docs conventions for LLMs.

Memory search

The sibling tool for a Qor’s own memory.