pippit-skill

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

• Third-party content exposure detected (high risk: 0.80). The skill polls Pippit agent threads and automatically downloads result URLs returned by those runs (see SKILL.md steps 5–7 and scripts/nest/get_thread.py together with scripts/nest/download_results.py and _common.iter_result_urls), meaning it ingests arbitrary external URLs and Pippit-provided messages as part of its workflow and uses those artifacts/messages to decide which assets to publish.

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

• Potentially malicious external URL detected (high risk: 0.70). The skill makes runtime API calls to https://www.pippit.ai (PIPPIT_BASE) and parses the returned agent messages/interactions which directly drive local behavior (e.g., requires_action prompts, download URLs), and the skill relies on that external content for core operation.