doc-tspec-validator

Pass

Audited by Gen Agent Trust Hub on Apr 1, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill's functionality is limited to validating local documentation files against predefined schemas. No suspicious behaviors such as obfuscation, exfiltration, or credential harvesting were identified.\n- [COMMAND_EXECUTION]: The skill defines commands to execute local validation scripts (e.g., python ai_dev_ssd_flow/10_TSPEC/scripts/validate_utest.py and bash ai_dev_ssd_flow/10_TSPEC/scripts/validate_all_tspec.sh). These operations are confined to the project's repository and represent standard development environment tasks.\n- [DATA_INGESTION]: (Category 8 surface analysis) The skill processes markdown files located in the docs/10_TSPEC/ directory. While these files are untrusted inputs, the validation and 'auto-fix' logic is implemented in external scripts rather than via unconstrained LLM processing, which mitigates the risk of indirect prompt injection.\n
  • Ingestion points: Markdown documentation files within the docs/10_TSPEC/ project path.\n
  • Boundary markers: The instructions do not specify explicit delimiters to separate document content from agent instructions.\n
  • Capability inventory: The skill can execute Python and Bash scripts which have the capability to read and modify local documentation files when the --auto-fix flag is applied.\n
  • Sanitization: Validation is performed by dedicated scripts checking against a fixed schema; no specific sanitization of document text within the prompt instructions is documented.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 1, 2026, 06:41 PM
Security Audit — agent-trust-hub — doc-tspec-validator