qa-docs-coverage
Pass
Audited by Gen Agent Trust Hub on Oct 6, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The script
scripts/docs_freshness_report.pyinvokes thegitbinary viasubprocess.check_outputto retrieve commit timestamps. The call uses a list of arguments to prevent shell injection and validates that all target paths exist on the filesystem before processing them.\n- [EXTERNAL_DOWNLOADS]: Thescripts/check_external_links.pyutility performs HTTPHEADandGETrequests usingurllib.requestto verify documentation links. It includes safeguards like host allow/skip lists and configurable timeouts to control network activity.\n- [INDIRECT_PROMPT_INJECTION]: The auditing scripts ingest content and metadata from repository files, representing a surface for indirect prompt injection where content within the documentation could attempt to influence the agent.\n - Ingestion points: Documentation files and YAML metadata processed by
check_external_links.py,check_local_links.py, anddocs_freshness_report.py.\n - Boundary markers: Standard Markdown and YAML delimiters are present, providing basic structure for the ingested data.\n
- Capability inventory: The skill can execute local
gitcommands (scripts/docs_freshness_report.py) and perform network status checks (scripts/check_external_links.py).\n - Sanitization: The
docs_freshness_report.pyscript validates that paths extracted from untrusted metadata actually exist on the disk usingp.exists()prior to their use in shell commands, mitigating potential injection risks.
Audit Metadata