reuse-first

Pass

Audited by Gen Agent Trust Hub on Sep 5, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill provides various shell scripts that use git, grep, xargs, sed, and awk to scan the project directory for duplicate literals and repeated phrases in code comments.
  • [EXTERNAL_DOWNLOADS]: Recommends the execution of npx jscpd@5, which fetches and runs a tool from the official npm registry to perform structural code analysis.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from the local repository (source code and comments) to identify duplication, creating a minor injection surface.
  • Ingestion points: Local codebase, specifically string literals and developer comments parsed via git diff and grep (SKILL.md).
  • Boundary markers: None present in the shell command pipelines.
  • Capability inventory: The skill uses git, grep, and a package runner (npx) to analyze and output findings (SKILL.md).
  • Sanitization: Extracted text from comments is processed and displayed without explicit escaping or filtering.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 5, 2026, 09:35 AM
Security Audit — agent-trust-hub — reuse-first