seo-geo-claude-skills
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCREDENTIALS_UNSAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSMETADATA_POISONING
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to fetch and analyze external URLs and competitor content (e.g., via
/aaron:auditand/aaron:compete), creating a surface for indirect prompt injection attacks where malicious instructions hidden in search results or audited pages could influence agent behavior. - Ingestion points: Processes user-supplied URLs and external web content in research and optimization phases.
- Boundary markers: No explicit boundary markers or warnings to ignore instructions within ingested content are provided in the skill documentation.
- Capability inventory: The skill can read from and write to a local
memory/directory and execute shell scripts. - Sanitization: No sanitization or filtering of external content is specified.- [CREDENTIALS_UNSAFE]: The documentation references the environment variable
GSC_CREDENTIALS_PATHfor Google Search Console, pointing to a sensitive file path containing credentials.- [COMMAND_EXECUTION]: The skill includes instructions to execute a local shell script (bash scripts/recover-retired-warm.sh) for recovering archived state, representing a command execution vector.- [EXTERNAL_DOWNLOADS]: Installation instructions direct users to download and install packages from a GitHub repository (aaron-he-zhu/seo-geo-claude-skills) which is not identified as a trusted source.- [METADATA_POISONING]: There is a discrepancy between the stated author (ara.so/aaron-he-zhu) in the file content and the expected vendor (reason-machines), which may indicate misleading metadata.
Audit Metadata