audit-my-app

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

• Third-party content exposure detected (high risk: 0.90). The skill's Phase 3 "Check & Suggest Skills from skills.sh" workflow explicitly uses WebSearch to query skills.sh and instructs the agent to curl/download AGENTS.md from raw.githubusercontent.com (third-party GitHub repos) and then act on those results, meaning it fetches and ingests untrusted public content that can influence installation decisions and subsequent agent behavior.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly instructs at runtime to download and install remote agent definitions and installers via raw.githubusercontent.com (e.g., https://raw.githubusercontent.com/{publisher}/{repo}/main/skills/{skill-name}/AGENTS.md and the installer https://raw.githubusercontent.com/brainit-consulting/DreamForgeSoftwareAgentSkills/main/install.sh), which would inject remote AGENTS.md content that controls agent prompts or (in the curl | bash case) execute fetched code, meeting the flagging criteria.