The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes data returned by the openspec CLI, including dynamic fields such as context and operationGuidance. These fields represent an ingestion point for untrusted data which could contain malicious instructions. The skill proactively addresses this risk by providing the agent with strict guardrails: instructing it to treat the data as advisory prompt-level input, to maintain logical separation from CLI-controlled state, and to report conflicts rather than executing conflicting instructions.
[COMMAND_EXECUTION]: The skill uses openspec CLI commands (e.g., openspec status, openspec instructions) to interact with a local or remote specification store. These operations are focused on project management and implementation tracking, utilizing a restricted set of tools within the intended scope of the skill's functionality.

openspec-apply-change