agent-skills-manager

CRITICAL E005: Suspicious download URL detected in skill instructions.

• Suspicious download URL detected (high risk: 0.70). The URL is a direct raw GitHub link to an install.sh script and the skill prompt instructs using curl | bash — a high-risk pattern because it executes remote shell code from a third‑party repo (which may be legitimate but could be compromised or malicious), so it should be treated as suspicious until the repository and script are reviewed and verified.

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

• Third-party content exposure detected (high risk: 0.90). The SKILL.md explicitly instructs fetching/installing skills and the bootstrap installer from public locations (e.g., curl to raw.githubusercontent.com and corvus commands that populate ~/.corvus/repo and write rules into .claude/.codex/.cursor rule directories), so the agent would ingest and apply untrusted third‑party skill/rule content that can change its behavior.

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

• Potentially malicious external URL detected (high risk: 1.00). The skill's prerequisite runs a runtime install command that fetches and executes remote code via curl https://raw.githubusercontent.com/ravnhq/ai-toolkit/main/install.sh | bash, which directly executes external code and is a required dependency for the skill.