Return per-entry retrieval analytics: how often each entry was retrieved (impressions), opened (opens), followed by writes or logs, and whether it was stale when opened. Useful for understanding which memories are most actionable and which are frequently stale. Requires at least min_impressions r...
AI agents call memory_insights to retrieve information from Munin Memory without modifying anything — typically the context-gathering step in research, monitoring, and reporting workflows, before the agent takes action elsewhere.
memory_insights purely queries and analyzes existing memory entry metadata. It retrieves statistics about how memories were accessed but does not create, modify, delete, or execute any operations. This is a passive read operation with no side effects. Severity is low because analytics queries pose minimal risk even if misused by an agent—they cannot alter data or trigger external effects.
From the tool's definition Tool returns analytics on retrieval patterns (impressions, opens, staleness) without modifying data. Description states it 'return[s]' data and focuses on 'understanding' memory usage patterns. No create, update, delete, or execute operations are performed.
Attacks that exploit this kind of access
Return per-entry retrieval analytics: how often each entry was retrieved (impressions), opened (opens), followed by writes or logs, and whether it was stale when opened. Useful for understanding which memories are most actionable and which are frequently stale. Requires at least min_impressions retrieval events (default 3) to appear in results; when nothing clears the threshold the response carries an explanatory. It is categorised as a Read tool in the Munin Memory MCP Server, which means it retrieves data without modifying state.
Register the Munin Memory MCP server in PolicyLayer and add a rule for memory_insights: 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 Munin Memory. Nothing to install.
memory_insights 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 memory_insights 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 memory_insights. 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.
memory_insights is provided by the Munin Memory MCP server (magnus-gille/munin-memory). 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 →