The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes the openspec config command-line utility to manage global settings, including listing, getting, and setting configuration keys.
[PROMPT_INJECTION]: The skill is designed to inject user-defined project context and rules from openspec/config.yaml into the agent's instructions. This is a primary feature for improving response quality but creates a surface for indirect prompt injection if the project file is sourced from an untrusted repository.
Ingestion points: Data is read from openspec/config.yaml (the context and rules fields).
Boundary markers: Injected context is delimited using <project-context> and <project-rules> tags.
Capability inventory: The skill can execute local commands and modify project-level configuration files.
Sanitization: The instructions do not specify validation or sanitization of the injected text.

openspec-config