The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes a local command-line utility for state management during workflow execution.
Evidence: omh runtime record --skill cancel --harness goal-execution --status started in SKILL.md.
[INDIRECT_PROMPT_INJECTION]: The skill accepts user-supplied cancellation requests which are used to control the agent's workflow state.
Ingestion points: Required inputs include 'active workflow state' and 'cancellation intent' as specified in SKILL.md.
Boundary markers: The skill lacks formal delimiters but instructs the agent to verify actions and use the host's voice to maintain context integrity.
Capability inventory: The skill interacts with the local environment via the omh tool.
Sanitization: No explicit sanitization or validation mechanisms for the user input are described in the skill content.

omh-cancel