skills-marketplace-readiness

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

• Third-party content exposure detected (high risk: 1.00). The required workflow feeds outsider-authored free text into the agent’s LLM context via the held-out fixture repository copy (README/SKILL.md and other text files) at runtime: eval/run_harness.py:prepare_workspace() → container runs /opt/marketplace-agent/bin/agent with --repository pointed at that fixture.