skills/raddue/crucible/audit/Gen Agent Trust Hub

audit

Pass

Audited by Gen Agent Trust Hub on May 23, 2026

Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: The skill possesses a vulnerability surface for indirect prompt injection because it processes untrusted user content (the code and documents being audited) and interpolates this data into prompts for analysis agents.
  • Ingestion points: Untrusted artifact content is ingested in Phase 1 and Phase 2 of the audit process defined in SKILL.md.
  • Boundary markers: While the skill provides rigid instructions to its sub-agents via prompt templates, it does not employ explicit sanitization or delimiters to isolate untrusted artifact content from the agent instructions.
  • Capability inventory: The skill has the ability to read local repository files, use the gh CLI to interact with GitHub, and write status/scratch files to the local disk.
  • Sanitization: There is no evidence of sanitization or escaping of artifact content before it is interpolated into the analysis templates.
  • [EXTERNAL_DOWNLOADS]: The skill interacts with GitHub using the gh CLI tool to search for existing issues and file new findings. This is a neutral observation of a connection to a well-known developer service.
  • [COMMAND_EXECUTION]: The orchestrator executes shell commands via the gh utility to manage project issues as part of its reporting phase (Phase 4).
Audit Metadata
Risk Level
SAFE
Analyzed
May 23, 2026, 03:28 PM
Security Audit — agent-trust-hub — audit