AI agents use sharpen_memory to create or update resources in Tages — usually the action step of a workflow, after the agent has gathered context. Every call changes real data in your Tages environment.
The tool rewrites/modifies an existing memory entry by reformatting it into imperative language. This is a reversible data modification (Write), not destructive since the preview/confirm flow suggests the original can be reviewed before overwriting. Severity is medium because misuse could corrupt stored architectural conventions or decisions that guide AI coding behavior across sessions.
From the tool's definition 'Rewrite a memory value' and 'Returns a before/after preview unless confirmed=true' indicate this tool modifies stored memory content when confirmed
Attacks that exploit this kind of access
Rewrite a memory value into imperative form (ALWAYS/NEVER/MUST/DO NOT) using Claude Haiku. Returns a before/after preview unless confirmed=true. It is categorised as a Write tool in the Tages MCP Server, which means it can create or modify data. Consider rate limits to prevent runaway writes.
Register the Tages MCP server in PolicyLayer and add a rule for sharpen_memory: 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 Tages. Nothing to install.
sharpen_memory 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 sharpen_memory 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 sharpen_memory. 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.
sharpen_memory is provided by the Tages MCP server (ryantlee25-droid/tages). 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 →