pylot-workers

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

• Third-party content exposure detected (medium risk: 0.30). The skill queues a user-supplied free-text prompt via the runtime endpoint ${PYLOT_API}/missions/${PYLOT_JOB_ID}/workers/${WID}/prompt, and then polls reads last_output back from the worker, so outsider-authored text is ingested through the prompt workflow.

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

• Potentially malicious external URL detected (high risk: 0.70). The skill issues runtime HTTP requests to the gateway endpoints (e.g., "${PYLOT_API}/missions/${PYLOT_JOB_ID}/workers" and related paths) which spawn and drive remote workers that execute the posted prompts, so the external ${PYLOT_API} service directly causes remote code execution.