skills/phuongnz/skills/explain-repo/Gen Agent Trust Hub

explain-repo

Pass

Audited by Gen Agent Trust Hub on Jul 5, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses standard CLI tools (git, gh) to perform repository reconnaissance, including fetching remote URLs and analyzing commit history. These operations are essential for the tool's primary function and are performed locally on the target repository.
  • [PROMPT_INJECTION]: The skill processes untrusted content from the target repository, creating a surface for indirect prompt injection. 1. Ingestion points: Repository source code and git metadata (Step 2 and 3). 2. Boundary markers: The skill does not explicitly define delimiters for untrusted code in its subagent instructions. 3. Capability inventory: File system writes and shell command execution (git/gh). 4. Sanitization: No explicit sanitization of repository content is mentioned before it is included in documentation templates. However, the 'refute' pass and 'cite or cut' discipline provide internal verification of generated content.
  • [SAFE]: The skill includes a custom, self-contained Markdown renderer script (assets/md.js) and HTML templates. Static analysis of the included code confirms it is benign and serves only to render documentation locally without external network dependencies.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 5, 2026, 01:11 PM
Security Audit — agent-trust-hub — explain-repo