AI agents call get_recent_errors to retrieve information from Pigeon without modifying anything — typically the context-gathering step in research, monitoring, and reporting workflows, before the agent takes action elsewhere.
This is a read-only query operation that retrieves previously captured browser console errors and warnings. It has no side effects, does not execute code, modify data, or trigger actions. The data returned is diagnostic information about past events, making it a low-severity Read operation appropriate for debugging contexts.
From the tool's definition Tool name 'get_recent_errors' and description 'Return buffered browser errors' indicate retrieval of data with no side effects. The tool reads and reports existing error logs without modifying or executing anything.
Attacks that exploit this kind of access
Return buffered browser errors, newest first: console.error/warn, uncaught. It is categorised as a Read tool in the Pigeon MCP Server, which means it retrieves data without modifying state.
Register the Pigeon MCP server in PolicyLayer and add a rule for get_recent_errors: 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 Pigeon. Nothing to install.
get_recent_errors 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 get_recent_errors 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 get_recent_errors. 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.
get_recent_errors is provided by the Pigeon MCP server (pepperonas/pigeon). 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 →