incident-postmortem-drafter

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

• Third-party content exposure detected (high risk: 0.75). The required workflow ingests a user-supplied JSON incident file (--input), and the script renders outsider-authored free text fields like timeline[].event, impact.customer/internal, summary, contributing_factors, went_well, went_poorly, and action_items directly into the Markdown that is then fed to the LLM context (via the agent’s LLM prompt construction), so any text originating from outsiders in that JSON becomes LLM-readable prose.