critique-to-dex

Pass

Audited by Gen Agent Trust Hub on Jul 13, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill processes untrusted input from critique reports or pasted text, creating an indirect prompt injection surface.\n
  • Ingestion points: Processes content from local markdown files, pasted text, and previous conversation turns as specified in the 'Inputs' section of SKILL.md.\n
  • Boundary markers: No explicit delimiters or 'ignore instructions' warnings are provided to separate data from instructions during processing.\n
  • Capability inventory: The skill performs file system modifications by executing the 'dex create' command.\n
  • Sanitization: There is no evidence of input validation or sanitization to prevent adversarial content within reports from hijacking the agent's flow.\n- [COMMAND_EXECUTION]: The skill executes local shell commands to manage project tasks through the dex CLI tool.\n
  • Evidence: Invokes 'dex dir' to verify workspace settings and 'dex create' to generate epics and tasks.\n
  • Safety Controls: A mandatory 'Preview and confirm' step is implemented (Step 4 in SKILL.md), requiring user verification before any files are written.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 13, 2026, 04:55 PM
Security Audit — agent-trust-hub — critique-to-dex