kelly-seo
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from Google Search Console, specifically search queries and page URLs, which could potentially contain malicious instructions. However, the risk is mitigated by a mandatory human-in-the-loop review process for all proposed actions and UI-level HTML escaping.\n
- Ingestion points: scripts/sync_gsc.mjs (imports Search Console data into Busabase).\n
- Boundary markers: Absent in the analysis instructions; the agent processes query text as part of its core logic.\n
- Capability inventory: Modifies local data in Busabase and proposes site edits (executed by the agent outside the app after approval).\n
- Sanitization: UI rendering in content/kelly-seo-app/app/app.js uses escapeHtml(), but the agent logic relies on LLM guardrails for text processing.\n- [EXTERNAL_DOWNLOADS]: The skill synchronizes search analytics and handles authentication via official Google API domains (googleapis.com). These are well-known services required for the skill's dashboard functionality.
Audit Metadata