RAG: natural-language search over APG pattern Markdown and example source text. Returns the most similar chunks with patternId (and example slug when applicable). Follow up with apg_get_pattern / apg_get_example for full docs. Requires a pre-built index from
AI agents call apg_semantic_search to retrieve information from Accessibility without modifying anything — typically the context-gathering step in research, monitoring, and reporting workflows, before the agent takes action elsewhere.
This is a read-only semantic search tool that searches an index and returns matching results. It has no side effects, does not modify data, does not execute code or commands, and does not delete or create resources. The tool's sole purpose is information retrieval, fitting squarely in the Read category with low severity.
From the tool's definition Tool performs 'natural-language search' and 'returns the most similar chunks'—it retrieves and queries data with no modification, deletion, or execution of external operations. It is explicitly a RAG (Retrieval-Augmented Generation) search function.
Attacks that exploit this kind of access
RAG: natural-language search over APG pattern Markdown and example source text. Returns the most similar chunks with patternId (and example slug when applicable). Follow up with apg_get_pattern / apg_get_example for full docs. Requires a pre-built index from. It is categorised as a Read tool in the Accessibility MCP Server, which means it retrieves data without modifying state.
Register the Accessibility MCP server in PolicyLayer and add a rule for apg_semantic_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 Accessibility. Nothing to install.
apg_semantic_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 apg_semantic_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 apg_semantic_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.
apg_semantic_search is provided by the Accessibility MCP server (nicolasgalvez/accessibility-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 →