The Agent Skills Directory

[COMMAND_EXECUTION]: The skill interacts with the local environment exclusively through the openspec CLI tool. Commands are used to create change directories, retrieve templated instructions, and track the status of planning artifacts. Tool usage is scoped to the openspec command prefix.
[INDIRECT_PROMPT_INJECTION]: The skill processes structured data returned by the openspec CLI, which could theoretically contain malicious instructions.
Ingestion points: The instruction, context, and rules fields retrieved from openspec instructions --json in SKILL.md.
Boundary markers: The skill explicitly instructs the agent to treat these fields as internal constraints and strictly forbids copying them into the final output files.
Capability inventory: Limited to writing documentation files to resolved paths and executing openspec subcommands.
Sanitization: The skill requires the agent to transform user-provided descriptions into kebab-case names before they are used as arguments in shell commands, providing protection against basic command injection.

openspec-propose