ADK Review / Local

Standalone task skill under the adk-review category router. Produces a findings-first self-review of local working-tree changes (unstaged, staged, or branch-vs-base).

When to use

• Reviewing uncommitted edits before staging.
• Reviewing staged changes before committing.
• Reviewing the current branch against its base before pushing.
• Catching issues that are cheaper to fix locally than on a remote PR.

When NOT to use

• The change is already on a remote PR -> adk-review-pr
• Reviewer comments already exist -> adk-review-feedback
• Multi-dimensional repo audit -> adk-audit-repo
• Doc-only review -> adk-docs-review

Inputs

Input	Required	Notes
<target>	yes	unstaged / staged / branch (default branch vs origin's tracking base or main)
<base>	optional	Base ref for branch mode (default: tracking branch or main)
<focus>	optional	correctness / security / performance / style / all (default)
<scope>	optional	Path filter
--auto	optional	Skip approval gates

Workflow

1. Confirm intent - restate target, base, focus, scope. Approval gate unless --auto.
2. Capture diff - run the appropriate git diff (unstaged / staged / <base>...HEAD). Confirm files changed and line counts.
3. Read code - read changed files in working-tree state, plus immediate dependencies and tests. Repo evidence over guessing.
4. Run dimension passes - same dimensions as adk-review-pr (correctness / security / performance / style / tests).
5. Tier findings - assign severity (Blocker / Critical / Should Have / May Have / Nitpick / Question).
6. Validate - reread each finding against the diff to confirm it is real and actionable in this branch.
7. Report - findings-first markdown ordered by severity; recommend the next move (fix locally now, or proceed to commit / PR).

Severity ladder

Label	Meaning
Blocker	Fix before pushing - bug, security hole, broken contract
Critical	Strongly recommended fix; would normally block a PR
Should Have	Improvement that meaningfully raises quality
May Have	Optional polish
Nitpick	Style or taste only
Question	Reviewer uncertain; needs clarification

Finding template

### [<Severity>] <One-line summary>
- **File**: `path/to/file.ts:LINE-LINE`
- **Issue**: <2-3 sentence explanation>
- **Evidence**: <quoted snippet>
- **Suggested change**: <concrete recommendation, code if useful>
- **Why this severity**: <one sentence>

Output format

## Local Review: <target> (<files> files, +<add> / -<del>)
- Base: <base ref> (for branch mode)
- Focus: <focus>

## Verdict
<ready-to-commit | needs-fixes | needs-discussion>

## Findings

### Blockers
<finding blocks>

### Critical
<finding blocks>

### Should Have
<finding blocks>

### May Have
<finding blocks>

### Nitpicks
<finding blocks>

### Questions
<finding blocks>

## Out of Scope
- <items explicitly not reviewed and why>

## Recommended Next Step
- <e.g. "fix Blockers in src/auth/", or "ready for `adk-publish-commit`">

Need more detail on any finding?

Pre-commit gates

If the repo has pre-commit hooks (pre-commit, husky, lefthook, etc.), the report must:

• Note whether they were run (the hook output goes through the dimension passes too).
• Distinguish hook failures (from the tooling) from review findings (from this skill).

Branch-mode specifics

• Default base is the tracking branch (git rev-parse --abbrev-ref --symbolic-full-name @{u}); fall back to origin/main then main.
• Use git diff <base>...HEAD (three dots) to compare merge bases, not the moving base ref.
• For multi-commit branches, also skim each commit message for inconsistencies.

Anti-patterns

• Reviewing the diff without reading the surrounding code.
• Posting nitpicks ahead of Blockers.
• Marking "ready to commit" while validation has not been run.
• Generic "looks good" without enumerating what was inspected.
• Treating linter complaints as Blockers automatically; classify by impact.

Examples

adk-review-local --target staged --focus correctness,security

adk-review-local --target branch --base origin/main --scope src/auth/

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 is the base for comparison (origin/<branch>, main, last commit, custom)? — How to pick: Default = upstream of current branch; fallback to main. Custom = pass an explicit base ref.
2. Scope: full diff or only staged changes? — How to pick: Default = full diff (catches unstaged forgotten work). Staged-only when preparing a commit.
3. Focus: correctness, security, performance, style, all? — How to pick: All by default; narrow when re-reviewing after fixes.

Default report: Verdict + severity-grouped findings + commit-readiness call.

Detailed report (on request or --verbose): Add: per-file diff summary, lint/typecheck output, test-coverage delta vs base, suggested patches.

Artifact: local-review-report — Markdown report. The working tree is unchanged.

Artifact path: .temp/reports/review-local-<slug>.md

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 is the base for comparison (origin/<branch>, main, last commit, custom)? — How to pick: Default = upstream of current branch; fallback to main. Custom = pass an explicit base ref.
2. Scope: full diff or only staged changes? — How to pick: Default = full diff (catches unstaged forgotten work). Staged-only when preparing a commit.
3. Focus: correctness, security, performance, style, all? — How to pick: All by default; narrow when re-reviewing after fixes.

Default vs detailed output

Default report: Verdict + severity-grouped findings + commit-readiness call.

Detailed report (on request or --verbose): Add: per-file diff summary, lint/typecheck output, test-coverage delta vs base, suggested patches.

Artifact: local-review-report — Markdown report. The working tree is unchanged.

Artifact path: .temp/reports/review-local-<slug>.md

<!-- 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/examples.md	Example trigger phrases, invocation, and report shape.
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.
references/research-protocol.md	Source ordering, stop conditions, evidence buckets, citation discipline.
references/review-comment-format.md	Standard finding format with stable IDs and severities.

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