AI agents call what_broke to retrieve information from Sfgraph without modifying anything — typically the context-gathering step in research, monitoring, and reporting workflows, before the agent takes action elsewhere.
The tool is part of a local knowledge graph inspection suite. The incomplete description ('USE THIS for any') is uninformative, but the tool name and sibling tools (all read-only analysis utilities) suggest this queries for broken elements rather than executes arbitrary operations or modifies data.
From the tool's definition Tool name is 'what_broke' and description states 'USE THIS for any' — incomplete description provided. Based on naming convention and context as a knowledge graph analysis tool alongside 'explain_code', 'dead_code_audit', and 'analyze_field', this appears to…
Attacks that exploit this kind of access
USE THIS for any. It is categorised as a Read tool in the Sfgraph MCP Server, which means it retrieves data without modifying state.
Register the Sfgraph MCP server in PolicyLayer and add a rule for what_broke: 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 Sfgraph. Nothing to install.
what_broke 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 what_broke 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 what_broke. 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.
what_broke is provided by the Sfgraph MCP server (ryanstark24/sfgraph). PolicyLayer sits as a proxy in front of this server to enforce policies before tool calls reach the server.
what_broke is one line of Sfgraph's registry record.
The record carries the whole server: verified identity, auth posture, risk grade, every tool classified, recommended policy — re-checked continuously.
Teams ship this data inside their own products. See what a licence covers →