The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes local commands to manage the OpenSpec workflow. Evidence includes openspec --version, git log, mkdir, and CLI-specific commands like openspec new change and openspec archive.
[INDIRECT_PROMPT_INJECTION]: The skill has a surface for indirect prompt injection by scanning the codebase for TODO and FIXME comments to suggest tasks. However, these suggestions are presented for user approval and do not involve immediate execution of untrusted instructions.
Ingestion points: Code files scanned for task selection in Phase 2.
Boundary markers: Absent.
Capability inventory: File system write operations and execution of the openspec CLI.
Sanitization: Content is summarized for user selection.

openspec-onboard