convex-self-heal

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

• Third-party content exposure detected (medium risk: 0.30). In the required workflow, prod error details are first ingested from the user’s own “sentinel” table via the MCP run-once-query/monitor event (steps 2–3) and then the agent uses that evidence (including the reproducing input) to triage/root-cause and certify fixes (steps 4–7), so outsider-authored free text is read at runtime if it was submitted into prod errors by a third party.