using-superpowers

To ensure an agent invokes the correct skill before replying, you must enforce a workflow that calls the Skill tool at the start of every conversation to verify applicable skills and set the correct operating mode before generating any user-facing output.

Skill invocation workflow is the process of checking for relevant skills before generating a response. It is needed to align agent behavior with user instructions, apply guardrails, and delegate tasks correctly from the very beginning of an interaction.

Yes, you can override agent behavior. The system respects user directives first, then Superpowers skills, and finally the default system prompt to determine the response strategy, allowing behavior overrides when a relevant skill exists.

Yes, the agent must invoke a skill before asking clarifying questions. The skill assessment happens before any user-facing reply, ensuring that even initial clarifications follow the correct operating mode and guardrails.

The limitation is that it requires a defined skill to exist for the invocation to be effective; without frontmatter presence, exact skill naming, and a defined workflow, the agent cannot automatically activate the relevant guidance.