Natural-language conceptual search across ALL Engrams entity types (decisions, patterns, progress, custom data) using vector embeddings. Use when you have a conceptual query that keyword search would miss (e.g.,
AI agents call semantic_search_engrams to retrieve information from Engrams without modifying anything — typically the context-gathering step in research, monitoring, and reporting workflows, before the agent takes action elsewhere.
This tool retrieves and queries project memory data but does not modify, delete, or execute actions. It is a pure read operation that returns search results. The low severity reflects that unauthorized access to project context could be sensitive but lacks the blast radius of write, execute, or destructive operations. Confidence is high because the description clearly indicates search/retrieval semantics.
From the tool's definition Tool performs 'Natural-language conceptual search across ALL Engrams entity types' using 'vector embeddings'. The verb is 'search', which is a read operation with no side effects. The description explicitly frames this as a query tool.
Documented attack patterns abuse exactly the kind of access semantic_search_engrams gives an agent:
PolicyLayer is an MCP gateway — it sits between your AI agents and Engrams, and nothing reaches the server without passing your rules. This is the rule we recommend for semantic_search_engrams:
{
"version": "1",
"default": "deny",
"tools": {
"semantic_search_engrams": {}
}
}semantic_search_engrams is read-only, so it stays allowed — but everything else on the server is denied unless you say otherwise.
Free to start. No card required.
Natural-language conceptual search across ALL Engrams entity types (decisions, patterns, progress, custom data) using vector embeddings. Use when you have a conceptual query that keyword search would miss (e.g.,. It is categorised as a Read tool in the Engrams MCP Server, which means it retrieves data without modifying state.
Register the Engrams MCP server in PolicyLayer and add a rule for semantic_search_engrams: 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 Engrams. Nothing to install.
semantic_search_engrams 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 semantic_search_engrams 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 semantic_search_engrams. 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.
semantic_search_engrams is provided by the Engrams MCP server (stevebrownlee/engrams). PolicyLayer sits as a proxy in front of this server to enforce policies before tool calls reach the server.
Start from Engrams, add the rest of your stack, and see everything your agents can call. Then put policy on all of it.
Free to start. No card required.
42 Engrams tools catalogued and risk-classified — across an index of 43,000+ MCP servers.