seo-google
Pass
Audited by Gen Agent Trust Hub on Apr 6, 2026
Risk Level: SAFEPROMPT_INJECTIONNO_CODECOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [PROMPT_INJECTION]: The skill processes data from untrusted external sources, creating a potential surface for indirect prompt injection attacks.\n
- Ingestion points: Website content analyzed via the NLP tool (
scripts/nlp_analyze.py) and PageSpeed tool (scripts/pagespeed_check.py), and user-generated YouTube comments fetched via the video details tool (scripts/youtube_search.py).\n - Boundary markers: Absent. The instructions lack explicit markers or formatting rules to differentiate between fetched external data and core agent instructions.\n
- Capability inventory: The skill executes local Python scripts (
scripts/*.py) and performs network requests to various Google service endpoints.\n - Sanitization: Absent. No sanitization or validation mechanisms for external content are documented in the provided markdown instructions.\n- [NO_CODE]: While the skill's instructions describe the use of 11 Python scripts (e.g.,
scripts/google_auth.py,scripts/pagespeed_check.py), these script files were not part of the analyzed skill package. The overall security profile depends on the logic contained within these missing files.\n- [COMMAND_EXECUTION]: The skill performs shell command execution to run its internal Python scripts for SEO data collection. This is consistent with its stated purpose, provided the scripts are trustworthy and correctly handle user-supplied arguments like URLs.\n- [EXTERNAL_DOWNLOADS]: The skill facilitates data retrieval from official Google API domains (googleapis.com) to retrieve performance and search analytics data. These are established, well-known technology services and are documented neutrally.
Audit Metadata