The Agent Skills Directory

[COMMAND_EXECUTION]: The skill routinely invokes the openspec CLI to create changes, check status, and retrieve instructions (e.g., openspec new change, openspec status --json, openspec instructions). These commands are used as intended by the vendor to manage the development lifecycle.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from project files and CLI outputs to guide its behavior, which is an inherent risk in automated specification tools.
- Ingestion points: The skill reads GLOSSARY.md, SPEC.md files, and the JSON results from openspec status and openspec instructions (which includes context, rules, and instruction fields).
- Boundary markers: The instructions do not define specific delimiters to separate user-provided or CLI-provided data from the agent's core instructions, relying on the LLM to differentiate between guidance and content.
- Capability inventory: The agent has the ability to execute shell commands via the openspec CLI and write new files to the file system (e.g., creating specs at openspec/changes/<name>/specs/).
- Sanitization: There is no evidence of sanitization or validation performed on the fields retrieved from CLI JSON output before they are used to influence the generation of new artifact files.

openspec-ff-change