openspec-explore
Pass
Audited by Gen Agent Trust Hub on Oct 15, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute a CLI command to gather project context and active change status.
- Evidence: Executes
openspec list --jsonat the start of the session to check for existing changes and schemas. - [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from the user's codebase and project-specific artifacts, which could contain instructions designed to influence the agent.
- Ingestion points: Reads arbitrary files from the codebase, search results, and OpenSpec artifacts like
proposal.md,design.md, andspec.md. - Boundary markers: The skill does not employ specific delimiters or instruction-ignore blocks when reading external file content into the prompt.
- Capability inventory: The agent has the capability to read files, search the directory structure, and write/modify OpenSpec-specific markdown artifacts.
- Sanitization: No evidence of sanitization, filtering, or validation of the content read from external files was found.
Audit Metadata