openspec-propose
Pass
Audited by Gen Agent Trust Hub on Sep 20, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes various openspec CLI commands to manage project state and generate files. It incorporates security best practices by instructing the agent to wrap user-provided inputs like change names and store IDs in double quotes to prevent command injection in the shell environment.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests user building requests to generate planning artifacts, creating a surface for indirect prompt injection. This is mitigated by explicit boundary markers that limit the agent to planning only, instructions to derive safe kebab-case identifiers, and a requirement for a separate workflow before any implementation or code modification occurs.
Audit Metadata