ssot-reconcile

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

• Third-party content exposure detected (high risk: 0.90). The skill explicitly fetches and ingests user-generated records from Aitable via aitableFetch (SHEETS: Clients, Brands, Session_Handoffs, etc.) and also calls the Stripe API, and those third-party fields are parsed and directly influence findings and automated actions (e.g., aitableDelete/patch and DB updates in runFull/runFix), so untrusted external content can change agent behavior.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill calls the aitableFetch endpoints (e.g. https://aitable.ai/fusion/v1/datasheets/{sheetId}/records) at runtime and injects the fetched Aitable records into the SessionStart system message, meaning remote Aitable content can directly influence the agent's context/prompts and the script requires AITABLE_API_KEY (exits if missing), so this is a runtime external dependency that controls agent input.