standard-site

Pass

Audited by Gen Agent Trust Hub on Aug 24, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill includes a local validation script (scripts/validate.py) and provides instructions for the agent to run it locally using the uv package manager to check record schemas.
  • [SAFE]: The validation script utilizes the regex library from the official PyPI registry for advanced text processing and grapheme counting, which is a standard and safe dependency.
  • [PROMPT_INJECTION]: The skill processes external content like RSS or Markdown to map it to lexicons, which represents an indirect prompt injection surface. This is mitigated by a schema validator that enforces strict constraints.
  • Ingestion points: Record data passed to scripts/validate.py via file input or stdin during the conversion workflow.
  • Boundary markers: The skill relies on structured AT Protocol lexicon schemas to define clear field boundaries and data types.
  • Capability inventory: The skill's scripts are restricted to local data transformation and validation without network access or privileged operations.
  • Sanitization: The validation logic enforces strict data types, maximum character lengths, and pattern matching for all record fields.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 24, 2026, 05:03 PM
Security Audit — agent-trust-hub — standard-site