debug-sentry-monitor

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

• Third-party content exposure detected (high risk: 0.75). Yes: the workflow uses runtime-fetched outsider Sentry content via MCP calls like search_issues/get_sentry_resource (issue titles, stacktraces, breadcrumbs, tags, user/event data) which are then ingested into the agent’s LLM context for triage and root-cause analysis.

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

• Potentially malicious external URL detected (high risk: 0.70). The skill issues a runtime CallMcpTool to scrape external documentation at "https://docs.sentry.io/platforms//" (references/details.md) which fetches remote content that would be injected into the agent's context and can directly influence prompts/decisions.