openspec-explore
Pass
Audited by Gen Agent Trust Hub on Sep 20, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests codebase data and configuration to assist in thinking and artifact generation. Ingestion points: Existing codebase files and project-specific configuration (openspec/config.yaml). Boundary markers: Instructions explicitly state that project context and rules should be treated as constraints and not reproduced in output. Capability inventory: Access to the openspec CLI for project state management and markdown artifact generation; no application code execution or writing tools. Sanitization: Strict logical guardrails enforce a thinking-only stance that prohibits implementation.
- [COMMAND_EXECUTION]: The skill utilizes the openspec CLI tool for managing change proposals and project status. These operations are restricted to the openspec namespace as defined in the platform configuration.
Audit Metadata