voice-baseline

Pass

Audited by Gen Agent Trust Hub on Sep 15, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes user-provided writing samples to generate style instructions for an agent. While this establishes an ingestion surface for external content that could theoretically contain instructions, the risk is negligible as the script performs only statistical counts of words and sentence shapes.
  • Ingestion points: The scripts/census.py script reads Markdown files from a user-specified directory and a draft file.
  • Boundary markers: No specific delimiters or boundary markers are used in the generated baseline report.
  • Capability inventory: The skill involves file system read and write operations, and the resulting baseline file is intended to influence agent behavior.
  • Sanitization: The script uses regular expressions to strip YAML frontmatter and Markdown code blocks from the input text before processing.
  • [SAFE]: The script uses only Python standard libraries and performs simple file I/O operations consistent with its stated purpose. No malicious code, obfuscation, or unauthorized access patterns were detected.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 15, 2026, 10:20 AM
Security Audit — agent-trust-hub — voice-baseline