feat: split review commands — local-review (renamed) + new whole-codebase code-review

CHANGELOG.md

@@ -10,7 +10,8 @@ Please choose versions by [Semantic Versioning](http://semver.org/).
 
 ## Unreleased
 
-- docs: add `docs/three-command-review-split.md` — design note locking the proposed shape for splitting the review commands into three distinct scopes (`/coding:pr-review` remote diff, `/coding:local-review` local pre-commit diff (renamed from current `/coding:code-review`), new `/coding:code-review` whole-codebase audit). No behavior change in this PR; the doc proposes a 5-PR migration plan (rename → deprecation stub → new behavior → flip deprecation) and the three mechanisms (severity filter, rule-id dedup, baseline file `.code-review-baseline.yaml`) that make whole-codebase review useful without drowning in pre-existing tech debt. Optional `golangci-lint` passthrough for Go projects.
+- docs: add `docs/three-command-review-split.md` — design note locking the proposed shape for splitting the review commands into three distinct scopes (`/coding:pr-review` remote diff, `/coding:local-review` local pre-commit diff (renamed from current `/coding:code-review`), new `/coding:code-review` whole-codebase audit). The doc pins defaults for baseline-file location (`.code-review-baseline.yaml` at repo root, `--baseline-path` override), monorepo handling (one baseline per repo by default; per-subproject via override), `--refresh-baseline` clean-tree requirement, and `.gitignore` semantics (respected; no separate `.claude-ignore` introduced).
+- **feat: split review commands into 3 distinct scopes.** Renames current `/coding:code-review` → `/coding:local-review` (preserving its diff-vs-`HEAD~1` semantics for pre-commit local checks), AND introduces a brand-new `/coding:code-review` that scans the WHOLE codebase via `git ls-files`. New command ships with severity filter (default-on: Must Fix + Should Fix only; `--include-optional` to opt in), rule-id dedup at consolidation (N occurrences → 1 summary with sample sites), and baseline file (`.code-review-baseline.yaml` via `--refresh-baseline`) so subsequent runs only flag NEW findings (drift since last sweep) rather than the operator's full accepted tech-debt set. Design rationale: `docs/three-command-review-split.md`. 10 reference sites updated across `llms.txt`, `README.md`, `scenarios/*`, and 4 agent definitions to point at `/coding:local-review` (preserving old semantics). **Migration**: sharp behavior cutover on the `/coding:code-review` slot — operators previously relying on its diff-vs-`HEAD~1` behavior must move to `/coding:local-review`. The contrast pair `pr-review` (remote, branch vs target) / `local-review` (local, pre-commit) reads cleanly; `code-review` takes the unmarked whole-codebase slot.
 
 ## v0.25.0
 

README.md

@@ -7,7 +7,7 @@ Opinionated coding guidelines, quality review agents, and slash commands for Go
 
 ## Overview
 
-Writing consistent, idiomatic code across a large codebase is hard. This plugin bundles 50+ opinionated guides (Go architecture, error handling, testing, HTTP handlers, Python structure, Git workflow, documentation) together with specialized Claude Code agents that enforce them on your code. Install once, then run `/coding:code-review` or `/coding:pr-review` to review your work against the full ruleset.
+Writing consistent, idiomatic code across a large codebase is hard. This plugin bundles 50+ opinionated guides (Go architecture, error handling, testing, HTTP handlers, Python structure, Git workflow, documentation) together with specialized Claude Code agents that enforce them on your code. Install once, then run `/coding:local-review` or `/coding:pr-review` to review your work against the full ruleset.
 
 ## Requirements
 
@@ -35,11 +35,19 @@ Review your current branch against all guidelines:
 /coding:pr-review
 ```
 
-Review code in standard mode (7 agents) or full mode (14 agents):
+Review local uncommitted changes in selector mode (default — zero LLM spawns, in-session classify + adjudicate) or full mode (per-owner dispatch, concurrent agents):
 
 ```
-/coding:code-review standard
-/coding:code-review full
+/coding:local-review            # selector mode (default)
+/coding:local-review full       # per-owner dispatch
+```
+
+Audit the whole codebase (severity-filtered, baseline-aware):
+
+```
+/coding:code-review              # whole codebase, Must Fix + Should Fix only
+/coding:code-review --include-optional         # add Nice to Have
+/coding:code-review --refresh-baseline         # write current findings to .code-review-baseline.yaml
 ```
 
 Find relevant guides before starting work:
@@ -58,8 +66,9 @@ Commit with changelog and version bump:
 
 | Command | Description |
 |---------|-------------|
-| `/coding:pr-review` | Review pull request diff against standards |
-| `/coding:code-review [short\|standard\|full]` | Review code against guidelines (standard: 7 agents, full: 14) |
+| `/coding:pr-review` | Branch diff vs target — selector mode default; full mode = per-owner dispatch |
+| `/coding:local-review [short\|selector\|full]` | Local uncommitted/recent diff vs `HEAD~1` — selector mode default |
+| `/coding:code-review [--include-optional] [--refresh-baseline]` | Whole-codebase audit — severity-filtered (Must + Should) + baseline-aware (`.code-review-baseline.yaml`) |
 | `/coding:architecture-review [directory]` | Deep whole-codebase architectural review — top-down + dimensions, consolidated Must/Should/Could |
 | `/coding:check-guides "task"` | Find relevant guides before implementation |
 | `/coding:commit` | Git commit with changelog and versioning |
@@ -258,7 +267,7 @@ End-to-end acceptance walks for the doc-driven review pipeline, following the [d
 | # | Scenario | Validates |
 |---|---|---|
 | 001 | [toolchain-preflight](scenarios/001-toolchain-preflight.md) | Step 4.0 / Step 0 preflight blocks exit 1 with documented stderr when `ast-grep` / `sg` is absent from PATH |
-| 002 | [clean-pr-zero-findings](scenarios/002-clean-pr-zero-findings.md) | `/coding:code-review master` against a zero-violation diff produces empty Must Fix / Should Fix / Nice to Have (no LLM hallucination) |
+| 002 | [clean-pr-zero-findings](scenarios/002-clean-pr-zero-findings.md) | `/coding:local-review master` against a zero-violation diff produces empty Must Fix / Should Fix / Nice to Have (no LLM hallucination) |
 | 003 | [scaling-funnel-100-files](scenarios/003-scaling-funnel-100-files.md) | 100-file synthetic fixture: mechanical funnel ≤30s, distinct Owners ≤30 (structural ceiling on Step 4b LLM calls) |
 | 004 | [findings-exist-path](scenarios/004-findings-exist-path.md) | `/coding:pr-review` against the stable test PR [bborbe/maintainer#2](https://github.com/bborbe/maintainer/pull/2): Step 4a surfaces ≥4 findings, every Owner has an agent file, citation discipline holds |
 

agents/architecture-dimensions-assistant.md

@@ -29,7 +29,7 @@ The user (typically via `/coding:architecture-review`) gives you a target direct
 
 - **No structural review** — that's the sibling agents' job. If you find a layering violation, note it briefly and defer to the structural pass.
 - **No code modification** — read-only.
-- **No diff-scoped review** — whole-codebase only. For per-PR work, the user should use `/coding:code-review` instead.
+- **No diff-scoped review** — whole-codebase only. For per-PR work, the user should use `/coding:pr-review` (branch diff) or `/coding:local-review` (uncommitted) instead.
 - **No mechanical findings** — "function too long", "too many params", "name should be camelCase" are owned by linters and `go-quality-assistant` / `python-quality-assistant`. Your altitude is architectural.
 
 ## Output format

agents/go-architecture-assistant.md

@@ -422,7 +422,7 @@ Package boundary violations: <count>.
 
 ## Integration
 
-- Runs alongside `srp-checker` in `/coding:code-review full` — SRP checks unit-level, this checks cross-unit and extraction-quality
+- Runs alongside `srp-checker` in `/coding:local-review full` — SRP checks unit-level, this checks cross-unit and extraction-quality
 - Invoked directly via `/coding:audit-architecture [directory]`
 - Read-only — never edits code
 

agents/python-architecture-assistant.md

@@ -438,7 +438,7 @@ Import-time side effects: <count>.
 
 ## Integration
 
-- Runs alongside `python-quality-assistant` in `/coding:code-review full` — quality checks style/idioms, this checks cross-unit architecture and extract-quality
+- Runs alongside `python-quality-assistant` in `/coding:local-review full` — quality checks style/idioms, this checks cross-unit architecture and extract-quality
 - Invoked directly via `/coding:audit-architecture [directory]`
 - Read-only — never edits code