The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to process external data fields (context and operationGuidance) retrieved via the openspec CLI as behavior instructions. This creates a surface where instructions embedded in a project's metadata or CLI output could attempt to override the agent's logic. * Ingestion points: context and operationGuidance fields within the JSON output of openspec instructions apply (SKILL.md). * Boundary markers: The instructions explicitly tell the agent to keep these fields separate from the system state and to report any conflicts with built-in instructions. * Capability inventory: File reading (via contextFiles), file writing (updating task lists), and shell command execution using the openspec CLI. * Sanitization: No programmatic sanitization is defined; the skill relies on the agent's ability to identify and prioritize controlling workflow inputs.
[COMMAND_EXECUTION]: The skill executes multiple subcommands of the openspec CLI using the Bash tool. These include store list, status, list, and instructions, which are used to interact with local project configuration and state.

openspec-apply-change