openspec-proposal
Pass
Audited by Gen Agent Trust Hub on Feb 19, 2026
Risk Level: SAFE
Full Analysis
- [Data Exposure & Exfiltration] (SAFE): No sensitive data access or network activity detected. Operations are limited to the 'openspec/' folder.
- [Remote Code Execution] (SAFE): No remote code patterns identified.
- [Command Execution] (LOW): Uses Bash for directory creation. Input sanitization is performed via kebab-case conversion of user input.
- [Indirect Prompt Injection] (LOW): The skill processes user descriptions to generate documents. Evidence Chain: 1. Ingestion points: feature description; 2. Boundary markers: absent; 3. Capability inventory: Bash, Write, Read, Glob, Grep; 4. Sanitization: kebab-case conversion for IDs.
Audit Metadata