senpi-entrypoint

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

• Third-party content exposure detected (high risk: 0.90). The skill explicitly fetches and parses open/public third-party content (e.g., curl to https://raw.githubusercontent.com/.../catalog.json in Step 2, POSTs to the public Supabase MCP endpoint https://ypofdvbavcdgseguddey.supabase.co/functions/v1/mcp-server for leaderboard/strategy data, and the background checker scripts/check-skill-updates.py queries the GitHub API and raw SKILL.md files) and uses those untrusted sources to choose/deploy skills and drive agent actions, so external content can materially influence tool use and decisions.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly fetches and installs code and runtime content from the Senpi GitHub repo (e.g. https://github.com/Senpi-ai/senpi-skills) and raw files such as https://raw.githubusercontent.com/Senpi-ai/senpi-skills/refs/heads/main/catalog.json and SKILL.md via npx/curl and the check-skill-updates.py script at runtime, and those fetched files directly control agent messages, installation behavior, and may execute remote code (meets all three flagging criteria).