taste-engine

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill explicitly instructs configuring the MCP with the header "Authorization: Bearer ", which requires supplying an API key that the agent may need to insert verbatim into requests or commands, creating an exfiltration risk.

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

• Third-party content exposure detected (high risk: 0.85). The required workflow calls submit_brand(url) and then get_brand(submission_id), where the MCP extracts and returns website content/artifacts (e.g., html/css/screenshot-derived text) from a user-supplied URL at runtime, so outsider-authored free text from public web pages can be ingested into the agent’s LLM context via the MCP response.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill requires a runtime connection to the Taste Engine MCP at https://mcp.thetaste.ai/mcp (Authorization: Bearer ), which provides the submit_brand/get_brand tools and returns structured extraction data that is injected into the agent's context and directly governs its outputs.