AI agents call search_songs to retrieve information from Mcp Phish without modifying anything — typically the context-gathering step in research, monitoring, and reporting workflows, before the agent takes action elsewhere.
The tool searches songs from a music database (phish.net/phish.in APIs). This is a retrieval operation with no side effects. While the description is empty (lowering confidence slightly), the naming convention and context of read-only sibling tools strongly indicate this performs a search query only. No data creation, modification, deletion, or external execution occurs.
From the tool's definition Tool name 'search_songs' combined with sibling tools that are all query/retrieval operations (get_audio, get_reviews, get_show, get_song, search_audio_tracks, search_shows).
Attacks that exploit this kind of access
search_songs. It is categorised as a Read tool in the Mcp Phish MCP Server, which means it retrieves data without modifying state.
Register the Mcp Phish MCP server in PolicyLayer and add a rule for search_songs: 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 Mcp Phish. Nothing to install.
search_songs 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 search_songs 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 search_songs. 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.
search_songs is provided by the Mcp Phish MCP server (pete-builds/mcp-phish). 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 →