AI agents call find_reachable_sinks to retrieve information from Orihime without modifying anything — typically the context-gathering step in research, monitoring, and reporting workflows, before the agent takes action elsewhere.
The tool appears to search or retrieve reachable sinks (likely security-relevant endpoints or data flow destinations) from the indexed code graph. This is a retrieval/query operation with no side effects—it reads static analysis data rather than executing code, modifying repositories, or making financial transactions.
From the tool's definition Tool name 'find_reachable_sinks' indicates a query/search operation over the code knowledge graph. Sibling tools ('find_callees', 'find_callers', 'find_entry_points') are clearly Read operations for code traversal and analysis.
Attacks that exploit this kind of access
find_reachable_sinks. It is categorised as a Read tool in the Orihime MCP Server, which means it retrieves data without modifying state.
Register the Orihime MCP server in PolicyLayer and add a rule for find_reachable_sinks: 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 Orihime. Nothing to install.
find_reachable_sinks 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 find_reachable_sinks 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 find_reachable_sinks. 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.
find_reachable_sinks is provided by the Orihime MCP server (srinivasan-sundaresan95/orihime). 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 →