The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes external project files to inform its analysis, which introduces a surface for indirect prompt injection.
Ingestion points: The agent reads project configuration files (openspec/config.yaml, openspec/config.yml) and existing OpenSpec artifacts (e.g., design.md, spec.md, proposal.md) discovered via the openspec status and openspec list commands.
Boundary markers: The instructions include a specific directive to prevent unintended reproduction of untrusted content: "Ground your thinking in these. They are constraints for you to follow, not content to reproduce: do NOT copy them into the conversation or into any artifact you create."
Capability inventory: The skill is restricted by the platform to executing the openspec CLI via Bash. It can read project structure, manage change metadata, and write documentation artifacts to local project directories.
Sanitization: The skill relies on the large language model's adherence to instructional guardrails to prevent executing instructions embedded within the files it analyzes.

openspec-explore