load-mr-context

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

• Third-party content exposure detected (high risk: 0.85). The skill uses GitLab MCP tools at runtime to fetch an open merge request’s description, diffs, and review discussions (outsider-authored free text from other contributors) via get_merge_request, get_merge_request_diffs, and discussion/review retrieval, which is then summarized into the agent’s LLM context.