ralph-upgrade

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

• Third-party content exposure detected (high risk: 0.90). The SKILL.md workflow explicitly fetches/clones the public GitHub repo https://github.com/tolulawson/ralph-harness (steps 2–3) and requires the agent to read and follow UPGRADING.md as the authoritative upgrade workflow, so arbitrary repository content could directly influence tool use and actions.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly fetches/clones https://github.com/tolulawson/ralph-harness at runtime and runs scripts from the checked-out source (e.g., python3 scripts/check-upgrade-surface.py and python3 scripts/migrate-installed-runtime.py), which executes remote code and thus directly controls runtime behavior.