When to use Spec Kitty modes
Choose the lightest path that still fits the work. Spec Kitty offers several entry points: skip it entirely for trivial edits, dispatch a specialist for a short task, run a bounded adversarial review at an SDD checkpoint, or drive a full mission when acceptance criteria and structured review matter.
No Spec Kitty
When: Trivial edit / day-to-day velocity; human as primary orchestrator.
Why: Fixed mission overhead dominates small changes.
Use your editor and agent directly. Reserve Spec Kitty for work that benefits from tracked artifacts, governance, or review gates.
Ad-hoc specialist session
When: Short task needing a specialist profile; no full Spec->Review.
Why: Use spec-kitty dispatch; lighter than a mission.
See Start an ad-hoc specialist session for examples and command syntax.
Adversarial squad
When: Bounded multi-lens review at an SDD point-cut (post-spec / post-plan / post-tasks).
Why: Catch fakeability and scope issues without replacing the mission pipeline.
Run adversarial review when an artifact is ready for challenge but you are not starting a new mission. It complements, not replaces, Spec->Plan->Tasks->Implement->Review.
Full mission
When: Multi-step change (new capability or a bug fix) with acceptance + structured review.
Why: Spec->Plan->Tasks->Implement->Review. Parallel agents are a mission capability, not a separate mode.
Expect bookkeeping commits as the mission advances work packages. See Understanding Spec Kitty Missions and Multi-agent workflow for mission setup and parallel work.
Compare the four paths
Alternate visuals
Alternate stylized splash — decorative only. The four actual paths are No Spec Kitty, Ad-hoc specialist, Adversarial squad, and Full mission, compared above.
