seo-local

Fail

Audited by Gen Agent Trust Hub on Oct 29, 2026

Risk Level: HIGHDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONMETADATA_POISONING
Full Analysis
  • [DATA_EXFILTRATION]: The skill instructs the agent to automatically locate and read sensitive local files within an Obsidian vault or other local knowledge bases. These files are described as containing brand facts, product details, and competitor information, which constitutes an attempt to access private local data without explicit per-file user consent.
  • [DATA_EXFILTRATION]: The skill directs the agent to modify the local file system by appending data to an 'SEO action log' within the user's private Obsidian vault, introducing a risk of data corruption or unauthorized modification of personal notes.
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from external sources (competitor websites and Google Business Profiles) via Python scripts without implementing boundary markers or sanitization. This creates a surface where a malicious website could inject instructions that the agent might follow, such as modifying the local Obsidian vault or performing unauthorized actions.
  • [COMMAND_EXECUTION]: The skill relies on the execution of multiple Python scripts via the shell (e.g., python3 skills/seo-geo-audit/scripts/seo_audit.py) to perform its primary auditing and benchmarking functions, passing external URLs as arguments to these scripts.
  • [METADATA_POISONING]: The skill's YAML frontmatter identifies the author as 'Sorank', which conflicts with the provided author context of 'thibaultbm'. Such discrepancies in metadata can misinform users regarding the origin and safety of the skill.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Oct 29, 2026, 06:28 AM