deliverable-fidelity-pl

Pass

Audited by Gen Agent Trust Hub on Jul 14, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses the Bash tool to run a local Node.js script (scripts/fidelity-check.mjs) which performs mechanical verification of document content against a list of findings.
  • [PROMPT_INJECTION]: The skill processes untrusted 'deliverable' text which presents a surface for indirect prompt injection.
  • Ingestion points: The skill reads arbitrary text from a 'deliverable' field in an input JSON file.
  • Boundary markers: No specific delimiters or safety instructions are defined to separate the untrusted deliverable text from the agent's instructions during the LLM spot-check phase.
  • Capability inventory: The skill is restricted to Bash and Read tools and does not have network access or write permissions.
  • Sanitization: The mechanical check script performs text normalization and character cleaning, though semantic sanitization for the LLM evaluation step is not present.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 14, 2026, 08:27 AM
Security Audit — agent-trust-hub — deliverable-fidelity-pl