twg-engineering-work

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

• Third-party content exposure detected (high risk: 0.85). The skill’s runtime workflow (e.g., “fetch/context the Jira workitem”, “hydrate PR details, comments, tasks, pipeline status, and diff”, and “resolve exact PR URL/ID”) necessarily ingests outsider-authored free text from third-party systems like Jira issue bodies and PR descriptions/comments/diffs authored by other contributors, which the agent then places into the LLM context.