project-weekly-report

Pass

Audited by Gen Agent Trust Hub on Aug 9, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes git commands through a Python script (scripts/collect_git_work.py) using subprocess.run. The commands are constructed using argument lists, which prevents shell injection. The script accesses local repository data and configuration (user.name, user.email) to generate report evidence.
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it processes untrusted data from Git repositories.
  • Ingestion points: Commit subjects, bodies, and file paths are extracted from the repository and passed to the agent for summarization in SKILL.md.
  • Boundary markers: There are no explicit delimiters or instructions to the agent to treat the commit data as untrusted or to ignore embedded instructions.
  • Capability inventory: The skill can execute the collector script and generate a report, but does not have broader system or network access.
  • Sanitization: The data is formatted as JSON, but the textual content within commit messages is not filtered or sanitized for potential injection patterns.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 9, 2026, 06:07 AM
Security Audit — agent-trust-hub — project-weekly-report