report-poc-to-requester

Pass

Audited by Gen Agent Trust Hub on Nov 11, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [SAFE]: The skill is designed as a security-enhancing tool that sanitizes reports to prevent the leakage of internal data such as absolute file paths, emails, and secrets.
  • [SAFE]: The included Python validator script (scripts/check_summary.py) operates in a read-only manner and uses strict regular expressions to identify and flag forbidden information.
  • [COMMAND_EXECUTION]: The skill executes a local Python script (check_summary.py) to validate generated content. This execution is safe as it targets a specific provided script and is used for content verification purposes.
  • [COMMAND_EXECUTION]: The unit tests in scripts/test_check_summary.py use subprocess.run to verify the validator's logic. This implementation is safe as it uses fixed paths and list-based argument passing for testing purposes.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes data from validation folders and JSON record files.
  • Ingestion points: Reads PoC result reports, status logs, and evidence records (record.json).
  • Boundary markers: Relies on references/report-template.md and explicit instructions to produce output in a single code block.
  • Capability inventory: Capabilities include reading local filesystem files and executing the check_summary.py validator script via Python.
  • Sanitization: Employs scripts/check_summary.py which uses regex to detect and block absolute paths, emails, GUIDs, and placeholders in the final report output.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 11, 2026, 04:18 AM