The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests data from external sources via the openspec instructions apply command, which returns context and operationGuidance fields. This data is intended to influence agent behavior during implementation.
Ingestion points: SKILL.md (Step 3: Get apply instructions).
Boundary markers: The skill instructs the agent to treat these as "prompt-level behavior contracts" and "not enforceable checks," emphasizing that they do not permit bypassing a blocked state or replacing built-in instructions.
Capability inventory: The agent uses the Bash(openspec:*) tool to execute CLI commands and is instructed to modify project files to implement tasks and update checkboxes.
Sanitization: The instructions explicitly forbid copying context or operationGuidance verbatim into implementation files or planning artifacts unless specifically requested by the user.
[COMMAND_EXECUTION]: The skill instructs the agent to execute several shell commands using the openspec CLI to manage state and retrieve data, including openspec store list, openspec status, openspec instructions, and openspec list.

openspec-apply-change