headless-cli-agents

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFE
Full Analysis
  • [EXTERNAL_DOWNLOADS] (SAFE): The skill references the claude-agent-sdk and anyio libraries. The Agent SDK is maintained by anthropics, which is a recognized trusted organization.
  • [COMMAND_EXECUTION] (SAFE): Examples use standard tool invocations. The use of flags like --allowedTools to limit agent capabilities demonstrates a security-first approach to tool access.
  • [DATA_EXFILTRATION] (SAFE): No patterns of hardcoded credentials or access to sensitive file paths (e.g., SSH keys, AWS configs) were detected.
  • [PROMPT_INJECTION] (SAFE): The skill contains no instructions designed to bypass agent safety filters or extract system prompts.
  • [INDIRECT_PROMPT_INJECTION] (SAFE): The skill describes processing untrusted data (such as PR diffs), which is its primary intended use case. Evidence: 1. Ingestion points: PR diff analysis in bash script. 2. Boundary markers: None. 3. Capability inventory: Read, Grep, WebSearch. 4. Sanitization: None. Use of restricted tools serves as a mitigation.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 06:48 PM