The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the Bash tool to execute openspec CLI commands for listing changes and checking status, as well as standard filesystem commands like mkdir and mv to organize the archive directory.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from multiple sources, which constitutes an attack surface.
Ingestion points: Data is retrieved from the openspec instructions JSON output, tasks.md files, and specification documents.
Boundary markers: The agent is instructed to treat external guidance as advisory and to explicitly report any conflicts with the skill's hardcoded logic.
Capability inventory: The skill possesses the ability to modify the filesystem, run CLI tools, and invoke secondary workflows.
Sanitization: Explicit guardrails prevent the agent from verbatim copying of external instructions and require validation of instructions against the primary workflow steps.
[DYNAMIC_EXECUTION]: The skill dynamically invokes the openspec-sync-specs workflow based on the presence of delta specs, representing a form of runtime logic orchestration.

openspec-archive-change