agent-readiness

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

• Third-party content exposure detected (high risk: 0.75). The skill clones an outsider-owned GitHub repository at runtime (git clone ... https://github.com/${REPO}.git) and then reads its free-text files (e.g., README.md/AGENTS.md/CLAUDE.md and evidence from commands) into the agent’s LLM-driven report generation, creating an indirect prompt-injection risk from outsider-authored repo content.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill runtime fetches a git token from "$PYLOT_GATEWAY_URL/git-token?repo=$REPO" and then clones "https://x-access-token:${GH_TOKEN}@github.com/${REPO}.git", which causes arbitrary repository code to be fetched and potentially executed during installs/tests, so these runtime URLs directly enable execution of remote code.