AI agents call docs_search to retrieve information from Docs without modifying anything — typically the context-gathering step in research, monitoring, and reporting workflows, before the agent takes action elsewhere.
docs_search performs a query-only operation against local documentation files, returning metadata and previews without modifying, executing, or deleting any data. It is a pure information retrieval tool with no side effects, making it a Read category risk. The low severity reflects that misuse (e.g., extracting sensitive examples from docs) has minimal blast radius compared to destructive or executable tools.
From the tool's definition Tool is described as "Search docs for a substring" and "Returns hits with package, path, line number, preview". The server is explicitly for "read documentation directly from installed packages" with "no network calls".
Attacks that exploit this kind of access
Search docs for a substring. Returns hits with package, path, line number, preview, and the section heading the hit lives under. It is categorised as a Read tool in the Docs MCP Server, which means it retrieves data without modifying state.
Register the Docs MCP server in PolicyLayer and add a rule for docs_search: allow, deny, rate-limit, or require approval. Point your MCP client at the PolicyLayer proxy URL and the rule is enforced on every call, before it reaches Docs. Nothing to install.
docs_search is a Read tool with low risk. Read-only tools are generally safe to allow by default.
Yes. Add a rate_limit block to the docs_search rule in your PolicyLayer policy. For example, setting max: 10 and window: 60 limits the tool to 10 calls per minute. Rate limits are tracked per agent session and reset automatically.
Set action: deny in the PolicyLayer policy for docs_search. The AI agent will receive a policy violation error and cannot call the tool. You can also include a reason field to explain why the tool is blocked.
docs_search is provided by the Docs MCP server (particle-academy/docs-mcp). PolicyLayer sits as a proxy in front of this server to enforce policies before tool calls reach the server.
Every MCP server has a record like this.
Type a name, get the same breakdown: verified identity, auth posture, risk grade, capabilities, recommended policy.
Teams ship this data inside their own products. See what a licence covers →