agentic-harness-engineering

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

• Third-party content exposure detected (medium risk: 0.65). The required runtime workflow includes an “Analyze”/“Improve” loop that uses web search (SERPER_API_KEY) for the evolve_agent, so fetched public web pages/articles can be ingested as free-form text into the agent’s LLM context via the web-search tool path.

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

• Potentially malicious external URL detected (high risk: 0.90). The README includes an explicit remote code execution during setup—"curl -LsSf https://astral.sh/uv/install.sh | sh" (and a required git clone of https://github.com/china-qijizhifeng/agentic-harness-engineering.git)—which fetches and runs external code as a required installation/runtime dependency.