ADK Plan (Category Router)

Routes any "think before you build" intent to the right planning task. Activate one of the listed task skills below; do not plan directly from this router.

When to use this category

• A request still has real ambiguity about goal, scope, or approach.
• Multiple valid approaches exist and the choice has trade-offs.
• A spec, design doc, ADR, or implementation plan is the deliverable.
• Research must precede a decision because facts are not yet known.

When NOT to use this category

• Implementation after the direction is settled -> adk-build
• Reviewing existing code or PRs -> adk-review
• Writing user-facing or maintenance docs -> adk-docs
• Auditing an existing repo or site -> adk-audit

Shared workflow

Every task in this category honors the same five steps. The task skills tailor the content of each step.

1. Confirm intent - restate the question or goal, the desired confidence, the acceptable blast radius (surgical / bounded / transformative), and the deliverable. Approval gate unless --auto.
2. Gather - pull repo evidence first, then primary external sources. Mark each finding Verified / Inferred / Open.
3. Synthesize - draft options or the chosen direction in concise markdown.
4. Validate - check the artifact against the success criteria; fix gaps.
5. Report - lead with the recommendation; show evidence, open questions, and the recommended next skill (often adk-build).

Task selection

Task skill	Use when	Do NOT use when
adk-plan-brainstorm	Ambiguity to close, blast radius to choose, 2-3 options to weigh, direction not yet locked	The decision is already made - skip to spec or roadmap
adk-plan-research	A factual question must be answered (framework behavior, upstream API, library comparison)	The answer is in this repo or already obvious - skip research
adk-plan-spec	Need a written spec (PRD, RFC, functional or technical spec) before implementation	A short plan is enough - use roadmap
adk-plan-design	Need an architecture / system design / ADR / data model write-up	The work is feature-scale and needs only an implementation plan
adk-plan-roadmap	Direction and design are settled; need an ordered, scoped implementation plan with milestones	Direction is still unresolved - brainstorm first

Common chains

flowchart LR
    brainstorm[adk-plan-brainstorm] --> research[adk-plan-research]
    research --> spec[adk-plan-spec]
    spec --> design[adk-plan-design]
    design --> roadmap[adk-plan-roadmap]
    roadmap --> build[adk-build]
    brainstorm --> roadmap
    brainstorm --> design

You almost never need every task. Common chains:

• Quick fix with one open question: brainstorm -> roadmap -> adk-build.
• New feature with unknown library behavior: research -> spec -> roadmap -> adk-build.
• New subsystem: brainstorm -> design -> roadmap -> adk-build.

Activation

Once you have picked a task, load adk-plan-<task> and follow it. Each task skill is fully standalone - everything it needs is in its own folder.

Anti-patterns

• Brainstorming when the user already chose - just write the spec or roadmap.
• Researching forever - cap evidence gathering by the confidence target.
• Skipping straight to a roadmap when there is real ambiguity.
• Writing implementation steps inside this router. Always hand off to a task skill.

Clarifying questions (default-ask)

When running without --auto, the skill asks these questions in order, one at a time. Under --auto, the skill picks the safest option for each (see references/clarifying-questions.md) and reports the choices.

1. What's missing right now: a decision, a fact, a written spec, an architecture, or an ordered plan? — How to pick: Decision → brainstorm. Fact → research. Spec → spec. Architecture → design. Ordered plan → roadmap.
2. Is the direction already locked? — How to pick: Yes → skip brainstorm; go to spec/design/roadmap. No → brainstorm.

Default report: One line: chosen plan task + why.

Detailed report (on request or --verbose): Gap audit table (decision/fact/spec/design/plan) showing which are present and which are missing.

Artifact: plan-routing-decision — Inline message.

Artifact path: (none)

Clarifying questions (default-ask)

When running without --auto, the skill asks these questions in order, one at a time. Under --auto, the skill picks the safest option for each (see references/clarifying-questions.md) and reports the choices.

1. What's missing right now: a decision, a fact, a written spec, an architecture, or an ordered plan? — How to pick: Decision → brainstorm. Fact → research. Spec → spec. Architecture → design. Ordered plan → roadmap.
2. Is the direction already locked? — How to pick: Yes → skip brainstorm; go to spec/design/roadmap. No → brainstorm.

Default vs detailed output

Default report: One line: chosen plan task + why.

Detailed report (on request or --verbose): Gap audit table (decision/fact/spec/design/plan) showing which are present and which are missing.

Artifact: plan-routing-decision — Inline message.

Artifact path: (none)

<!-- adk:references:start -->

References shipped with this skill

These files live in references/ next to this SKILL.md. Read them when the skill activates; they are inlined here so the skill is fully self-contained (no cross-skill or shared sources).

File	Purpose
references/anti-patterns.md	Things to avoid when running this skill.
references/artifact-format.md	The deliverable's format and where it lives (.temp/ contract).
references/clarifying-questions.md	The default-ask questions for this skill, with how-to-pick rubrics.
references/constitution.md	Non-negotiable rules and working/communication discipline.
references/interaction-contract.md	Default-ask, explained-options, --auto contract every skill must follow.
references/output-format.md	Default vs detailed report shapes; severity labels; verbosity rules.
references/persona.md	The agent persona that drives this skill.

<!-- adk:references:end -->