darwinian-evolver

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

• Third-party content exposure detected (high risk: 0.75). Outsider free text is ingested into the LLM via the mutator/evaluator prompts that include failure_case.response (LLM output) and failure_case.phrase (test strings) inside scripts/parrot_openrouter.py, where _prompt_llm() sends the constructed prompt to OpenRouter/OpenAI; the failure_case.response originates from the provider’s model output (not authored by the operating user).

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

• Potentially malicious external URL detected (high risk: 0.90). The skill's install/run instructions explicitly git-clone and invoke the upstream repository https://github.com/imbue-ai/darwinian_evolver.git (clone then uv run), so remote code is fetched and executed and is required for the skill to run.