AI agents use fail_task to create or update resources in ATMcp — usually the action step of a workflow, after the agent has gathered context. Every call changes real data in your ATMcp environment.
This tool changes the status of a task in a multi-agent coordination system. While it transitions a task to a failed state, this is reversible—tasks can be re-queued or reset by administrative action. The effect is state modification of a work item, not irreversible deletion.
From the tool's definition Tool description states: 'Report a task failed. Re-queued (open) until max_attempts, then dead-lettered (failed).' This modifies task state from active to failed/dead-lettered, which is a reversible change in a task management system (can be reset or…
Attacks that exploit this kind of access
Report a task failed. Re-queued (open) until max_attempts, then dead-lettered (failed). It is categorised as a Write tool in the ATMcp MCP Server, which means it can create or modify data. Consider rate limits to prevent runaway writes.
Register the AT MCP server in PolicyLayer and add a rule for fail_task: 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 ATMcp. Nothing to install.
fail_task 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 fail_task 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 fail_task. 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.
fail_task is provided by the AT MCP server (midcheck/atmcp). 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 →