skill-bundle-curator

Pass

Audited by Gen Agent Trust Hub on Jul 5, 2026

Risk Level: SAFE
Full Analysis
  • [COMMAND_EXECUTION]: The script scripts/check.py uses subprocess.run to call local Python validation tools. These calls use statically defined or locally derived paths and do not incorporate external user input into the command string.
  • [DATA_EXFILTRATION]: The skill reads directory structures and the agent-bundle.json file to generate inventory reports. All data processing occurs locally, and no network operations or external data transmissions were found.
  • [PROMPT_INJECTION]: The skill ingests data from a JSON configuration file, which is a surface for indirect prompt injection. However, the SKILL.md includes explicit 'Critical Rules' instructing the agent to treat potential command strings as documentation only and to refuse live execution or installation.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 5, 2026, 02:05 AM
Security Audit — agent-trust-hub — skill-bundle-curator