implement-rfc

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

• Third-party content exposure detected (high risk: 0.95). Runtime path: the skill uses WebFetch to read a GitHub discussion URL (and then fetches the full thread and linked issues via gh issue view), which ingests outsider-authored free text (community comments/RFC content) into the agent’s LLM context.

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

• Potentially malicious external URL detected (high risk: 1.00). The skill fetches and follows content from GitHub at runtime (e.g. https://github.com/remix-run/react-router/discussions/ and via gh issue view <number> --repo remix-run/react-router), so external discussion text is required at runtime and directly controls the agent's instructions/implementation.