AI agents use reflect to create or update resources in Engram — usually the action step of a workflow, after the agent has gathered context. Every call changes real data in your Engram environment.
The reflect tool processes existing facts and generates new derived data (patterns, preferences, insights), writing new higher-order observations into the knowledge store. It is not purely reading since it creates new synthesized entries, and it is not destructive or financial. Write is the most appropriate category as it produces new stored artifacts from existing data.
From the tool's definition Synthesize higher-order observations from accumulated facts. Groups facts by shared entities and generates patterns, preferences, and insights.
Attacks that exploit this kind of access
Synthesize higher-order observations from accumulated facts. Groups facts by shared entities and generates patterns, preferences, and insights. It is categorised as a Write tool in the Engram MCP Server, which means it can create or modify data. Consider rate limits to prevent runaway writes.
Register the Engram MCP server in PolicyLayer and add a rule for reflect: 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 Engram. Nothing to install.
reflect is a Write tool with medium risk. Write tools should be rate-limited to prevent accidental bulk modifications.
Yes. Add a rate_limit block to the reflect 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 reflect. 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.
reflect is provided by the Engram MCP server (reallyartificial/engram). 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 →