The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data in the form of operational guidance, context, and specification rules retrieved via the openspec CLI and local files.
Ingestion points: Data is ingested from openspec instructions archive (fields: context, operationGuidance), openspec instructions specs (field: rules), tasks.md (task status), and delta specification files.
Boundary markers: The skill contains explicit instructions to keep external guidance separate from built-in logic, stating that guidance is "advisory and optional" and must not block the workflow. It directs the agent to report conflicts and prioritize "explicit user choices, resolved paths, CLI checks, and command contracts."
Capability inventory: The agent has access to the openspec CLI, filesystem operations (mkdir, mv), and the ability to execute the openspec-sync-specs workflow.
Sanitization: The instructions include a verification step after synchronization to ensure main specifications match expected states, and explicitly forbid copying guidance or rule text verbatim into output files.
[COMMAND_EXECUTION]: The skill relies on shell command execution to perform its primary functions.
Evidence: The workflow uses openspec commands (e.g., status, list, instructions, archive) to manage change states and uses standard shell utilities like mkdir -p and mv to manipulate the filesystem during the archiving process.

openspec-archive-change