fp-go-mcp

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (medium risk: 0.30). In the fp-go MCP server, outsider-authored free text can reach runtime LLM ingestion via tool-returned use_skill/search_examples/get_example content in response to user-supplied tool parameters (e.g., search_examples.query), even though the underlying data is embedded and offline.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The skill requires installing and running the external generator github.com/IBM/fp-go/gen/v2 (via "go get -tool github.com/IBM/fp-go/gen/v2" and then "go tool gen mcp"), which fetches and executes third-party code that provides the MCP skill content used to control assistant behavior.