@ps-neko/nekowork 0.1.0-alpha.4 → 0.1.0-alpha.6

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
package/CLAUDE.md CHANGED
@@ -8,7 +8,7 @@
8
8
 
9
9
  ## 자동 갱신 영역
10
10
 
11
- <!-- HARNESS:START version=0.1.0-alpha.4 -->
11
+ <!-- HARNESS:START version=0.1.0-alpha.6 -->
12
12
  <!-- 이 영역은 scripts/sync-claude-md.js 가 자동 갱신한다. 직접 편집 금지. -->
13
13
 
14
14
  ## 카탈로그 요약
@@ -17,8 +17,8 @@
17
17
  - skills: 10
18
18
  - commands: 1 (legacy compat)
19
19
  - hooks: 5 (gateguard-fact-force, config-protection, quality-gate, pre-bash-dispatcher, persistent-mode)
20
- - packs: core, quality, security, frontend, testing, release, enterprise
21
- - profiles: core, developer, security, product, quality, frontend, testing, research, full
20
+ - packs: core, builder, quality, security, frontend, testing, release, enterprise
21
+ - profiles: core, developer, builder, security, product, quality, frontend, testing, research, full
22
22
  - harnesses: claude, codex, cursor, gemini, opencode
23
23
 
24
24
  ## 에이전트 → 모델 매트릭스
package/README.md CHANGED
@@ -1,10 +1,10 @@
1
1
  # NEKOWORK
2
2
 
3
- Local-first AI development quality runtime for Claude Code, Codex CLI, and Gemini CLI.
3
+ Local-first AI development OS and quality runtime for Claude Code, Codex CLI, and Gemini CLI.
4
4
 
5
5
  [![harness-validate](https://github.com/Ps-Neko/NEKOWORK/actions/workflows/harness-validate.yml/badge.svg)](https://github.com/Ps-Neko/NEKOWORK/actions/workflows/harness-validate.yml)
6
6
 
7
- NEKOWORK prevents AI coding agents from shipping unverified changes.
7
+ NEKOWORK helps AI coding agents build quickly, then prevents unverified changes from shipping.
8
8
 
9
9
  It runs:
10
10
 
@@ -18,7 +18,7 @@ No auto-commit. No auto-push. No surprise deploy.
18
18
  Product principle:
19
19
 
20
20
  ```text
21
- NEKOWORK = Claude work -> Codex verification -> Human Gate
21
+ NEKOWORK = fast AI build -> Codex verification -> Human Gate -> explicit apply
22
22
  ```
23
23
 
24
24
  NEKOWORK is the product. HARNESS is the local runtime it packages: one source catalog, `agent.yaml`, projected into Claude Code, Codex CLI, Cursor, Gemini CLI, and OpenCode surfaces.
@@ -30,10 +30,12 @@ NEKOWORK is intentionally not a 100-agent pack. Every agent, skill, hook, profil
30
30
  3. produce auditable evidence,
31
31
  4. respect Human Gate.
32
32
 
33
- **Public alpha evidence:** 7 packs / 9 profiles / 36 components / 5 harness targets / 7 case-study flows / 252 tests / 0 moderate+ npm audit issues / fresh `npx @alpha` smoke
33
+ **Public alpha evidence:** 8 packs / 10 profiles / 36 components / 5 harness targets / 7 case-study flows / 263 tests / 0 moderate+ npm audit issues / fresh `npx @alpha` smoke
34
34
 
35
35
  NEKOWORK does not automatically commit, push, publish, deploy, or apply diffs. `apply` is explicit and requires verified ship-ready evidence.
36
36
 
37
+ **Latest alpha evidence:** [CI badge](https://github.com/Ps-Neko/NEKOWORK/actions/workflows/harness-validate.yml) / [npm package](https://www.npmjs.com/package/@ps-neko/nekowork) / [smoke transcript](docs/DEMO.md#one-minute-terminal-transcript) / [report artifact](docs/DEMO-REPORT.md)
38
+
37
39
  **One-minute demo:** [terminal transcript](docs/DEMO.md#one-minute-terminal-transcript) / [full report example](docs/DEMO-REPORT.md) / [alpha feedback](https://github.com/Ps-Neko/NEKOWORK/issues/new?template=alpha-feedback.yml) / [roadmap](docs/ROADMAP.md)
38
40
 
39
41
  ![NEKOWORK one-minute terminal demo](docs/assets/demo-terminal.svg)
@@ -44,9 +46,22 @@ Use the current npm alpha for the published health smoke:
44
46
 
45
47
  ```bash
46
48
  npx -y @ps-neko/nekowork@alpha check
49
+ npx -y @ps-neko/nekowork@alpha build "implement this safely" --mode fast --session first-build
50
+ npx -y @ps-neko/nekowork@alpha report --session latest
51
+ ```
52
+
53
+ Start with `build`. Drop down to `work`, `verify`, and `ship` only when you need phase-level control.
54
+
55
+ Use a source checkout for local development:
56
+
57
+ ```bash
58
+ node scripts/cli.js check
59
+ node scripts/cli.js build "implement this safely" --mode fast --session first-build
60
+ node scripts/cli.js report --session latest
61
+ node scripts/cli.js gate status --session latest
47
62
  ```
48
63
 
49
- Use a source checkout for the new simple command path:
64
+ Or use the decomposed beginner path directly:
50
65
 
51
66
  ```bash
52
67
  node scripts/cli.js check
@@ -55,7 +70,7 @@ node scripts/cli.js report --session first-run
55
70
  node scripts/cli.js gate status --session first-run
56
71
  ```
57
72
 
58
- The simple path maps to the full evidence loop: `check = doctor --quick`, and `run = work -> verify -> ship`.
73
+ The simple paths map to the evidence loop: `check = doctor --quick`, `build = mode presets over run`, and `run = work -> verify -> ship`.
59
74
 
60
75
  To add generated harness surfaces to another local repository:
61
76
 
@@ -107,13 +122,29 @@ Decision:
107
122
 
108
123
  Human Gate is the point where NEKOWORK stops being an autopilot and becomes an approval system.
109
124
 
125
+ ## Apply Preview
126
+
127
+ Before `apply`, NEKOWORK expects the human to inspect the evidence surface:
128
+
129
+ ```text
130
+ Session: first-work
131
+ Diff source: captured live-work diff
132
+ Files changed: 3
133
+ Verifier verdict: approve
134
+ Human gate: clear
135
+ Ship ready: true
136
+ Apply command: node scripts/cli.js apply --session first-work
137
+ ```
138
+
139
+ `apply` still does not commit, push, publish, deploy, or create a PR. It only applies the verified `SHIP_READY` diff when gates are clear and the target worktree is clean.
140
+
110
141
  ## Compared With Agent Packs
111
142
 
112
143
  | Tool pattern | Optimizes for | NEKOWORK optimizes for |
113
144
  |---|---|---|
114
145
  | Large Claude Code packs | More agents, commands, skills | Curated verification loop |
115
146
  | Team simulation | More specialist perspectives | Read-only team plus one executor |
116
- | Autopilot | Fast autonomous execution | Report, gate, explicit apply |
147
+ | Autopilot | Fast autonomous execution | `build` modes, report, gate, explicit apply |
117
148
  | Discipline workflows | Better development habits | Evidence-backed ship decision |
118
149
 
119
150
  ## When To Choose What
@@ -124,13 +155,16 @@ Human Gate is the point where NEKOWORK stops being an autopilot and becomes an a
124
155
  | Get the broadest Claude Code skill/command environment | Everything Claude Code |
125
156
  | Simulate startup team roles from planning to QA | GStack |
126
157
  | Run autonomous multi-agent execution | OMC |
158
+ | Use one local AI development runtime with safe build modes | NEKOWORK |
127
159
  | Verify AI changes, require human approval, then apply explicitly | NEKOWORK |
128
160
 
161
+ Use Superpowers, Everything Claude Code, GStack, or OMC to produce stronger AI work when they fit your workflow. Use NEKOWORK as the main runtime when you want build speed plus verification, gate, report, and explicit apply in one product.
162
+
129
163
  ## Three Paths
130
164
 
131
165
  Most users should start with the Beginner path. The other paths are for explicit phase control or legacy compatibility.
132
166
 
133
- 1. Beginner: `check -> run -> report -> gate`
167
+ 1. Beginner: `check -> build -> report -> gate`
134
168
  2. Advanced: `ask -> plan -> team -> work -> verify -> gate -> ship -> report -> apply`
135
169
  3. Legacy: `review` / `review-cycle`
136
170
 
@@ -140,9 +174,10 @@ NEKOWORK is for teams that want AI-assisted development without making the agent
140
174
 
141
175
  ## Status
142
176
 
143
- - Current repository version: `0.1.0-alpha.4`
177
+ - Current repository version: `0.1.0-alpha.6`
144
178
  - Current package name: `@ps-neko/nekowork`
145
- - Current npm alpha: `@ps-neko/nekowork@0.1.0-alpha.4`
179
+ - Published CLI names: `nekowork` and `harness`
180
+ - Current npm alpha: `@ps-neko/nekowork@0.1.0-alpha.6`
146
181
  - Supported install path today: npm alpha, clone, submodule, or local repository integration
147
182
  - Dist-tag note: use `@alpha` until a stable release; `latest` still points at the first alpha line
148
183
  - Default mode: mock providers, no API keys, no provider CLI calls
@@ -150,7 +185,7 @@ NEKOWORK is for teams that want AI-assisted development without making the agent
150
185
  Current local verification:
151
186
 
152
187
  - `npm run lint`: pass
153
- - `npm test`: 252 tests pass
188
+ - `npm test`: 263 tests pass
154
189
  - `npm audit --audit-level=moderate`: 0 vulnerabilities
155
190
  - `npm pack --dry-run --json`: pass
156
191
  - `npx -y @ps-neko/nekowork@alpha check`: pass with warnings only
@@ -179,21 +214,11 @@ Current local verification:
179
214
  | `release` | ship/no-ship evidence | pre-release checks |
180
215
  | `enterprise` | full catalog with all gates | high-control teams |
181
216
 
182
- ## Quick Start
183
-
184
- Requirements:
217
+ ## Quick Start Details
185
218
 
186
- - Node.js 22+
187
- - npm
188
- - git
189
-
190
- Fastest no-API demo:
191
-
192
- ```bash
193
- npx -y @ps-neko/nekowork@alpha check
194
- ```
219
+ Requirements: Node.js 22+, npm, and git.
195
220
 
196
- Repository demo:
221
+ For a repository-pinned local demo:
197
222
 
198
223
  ```bash
199
224
  git clone https://github.com/Ps-Neko/NEKOWORK.git harness
@@ -202,21 +227,7 @@ npm ci
202
227
  npm run demo:quick -- --cleanup
203
228
  ```
204
229
 
205
- This creates a disposable target project and runs `doctor -> run -> report -> gate status`. It uses mock providers and does not call Claude, Codex, Gemini, or paid APIs.
206
-
207
- Recommended path for most users:
208
-
209
- ```bash
210
- git clone https://github.com/Ps-Neko/NEKOWORK.git harness
211
- cd harness
212
- npm ci
213
- node scripts/cli.js check
214
- node scripts/cli.js run "implement, verify, and prepare ship readiness" --session first-run
215
- node scripts/cli.js report --session first-run
216
- node scripts/cli.js gate status --session first-run
217
- ```
218
-
219
- `run` executes `work -> verify -> ship`. `report` turns the session evidence into a readable `REPORT.md`. It does not apply by default. `apply` is always explicit and requires a verified `SHIP_READY` live-work diff.
230
+ This creates a disposable target project and runs `doctor -> build -> report -> gate status`. It uses mock providers and does not call Claude, Codex, Gemini, or paid APIs.
220
231
 
221
232
  To initialize another local repository with the published alpha:
222
233
 
@@ -225,20 +236,6 @@ cd /path/to/my-project
225
236
  npx -y @ps-neko/nekowork@alpha init --profile developer --project-root .
226
237
  ```
227
238
 
228
- Advanced path:
229
-
230
- ```text
231
- ask -> plan -> team -> work -> verify -> gate -> ship -> report -> apply
232
- ```
233
-
234
- Legacy compatibility smoke:
235
-
236
- ```bash
237
- node scripts/cli.js review "check the project setup" --no-ship --session first-smoke
238
- ```
239
-
240
- The default review path uses mock providers, so it does not need API keys or provider CLIs.
241
-
242
239
  For the fuller first-run guide, see [docs/QUICKSTART.md](docs/QUICKSTART.md).
243
240
 
244
241
  For the trust and recovery model, see [Safety Guarantees](docs/SAFETY-GUARANTEES.md), [Failure Modes](docs/FAILURE-MODES.md), [Trust Model](docs/TRUST-MODEL.md), and [Why Not Autopilot](docs/WHY-NOT-AUTOPILOT.md).
@@ -251,11 +248,11 @@ npm run demo:external
251
248
 
252
249
  To inspect small case-study targets, see [examples/trading-dashboard-mock](examples/trading-dashboard-mock), [examples/github-actions-hardening](examples/github-actions-hardening), [examples/quality-lifecycle-smoke](examples/quality-lifecycle-smoke), and [docs/case-studies](docs/case-studies). They demonstrate financial UI, CI workflow, quality lifecycle, npm package, auth parser, Python protocol library, and environment configuration flows while still preserving Codex verification, Human Gate policy, and explicit apply control.
253
250
 
254
- ## What You Get
251
+ ## Output Shape
255
252
 
256
253
  ```text
257
254
  doctor ... OK
258
- run workflow ... OK
255
+ build workflow ... OK
259
256
  report ... OK
260
257
  gate status ... OK
261
258
  Demo completed: verdict=approve_with_fixes, ship_ready=false, applied=false
@@ -268,25 +265,14 @@ Outputs are written under:
268
265
  .harness/state/sessions/<session-id>/REPORT.md
269
266
  ```
270
267
 
271
- ## Use It In Another Project
272
-
273
- Shortest npm alpha install shape:
274
-
275
- ```bash
276
- cd <target-project>
277
- npx -y @ps-neko/nekowork@alpha init --profile developer --project-root .
278
- npx -y @ps-neko/nekowork@alpha check --project-root .
279
- ```
280
-
281
- Repository-pinned install shape:
268
+ ## Repository-Pinned Install
282
269
 
283
270
  ```bash
284
271
  cd <target-project>
285
272
  git submodule add https://github.com/Ps-Neko/NEKOWORK.git .harness-tool
286
273
  node .harness-tool/scripts/portability/simulate-port.js . --profile developer --verbose
287
274
  node .harness-tool/scripts/install-apply.js --profile developer --project-root .
288
- node .harness-tool/scripts/cli.js doctor --project-root . --quick
289
- node .harness-tool/scripts/cli.js plan "first NEKOWORK smoke" --project-root .
275
+ node .harness-tool/scripts/cli.js check --project-root .
290
276
  ```
291
277
 
292
278
  The HARNESS tool root stays in `.harness-tool/`. Session state, generated harness files, and git work happen in the target project root.
@@ -327,6 +313,7 @@ The public alpha surface is intentionally small:
327
313
  - `ship`: produce a ship/no-ship readiness handoff after Codex verification
328
314
  - `apply`: apply a verified `SHIP_READY` live-work diff to the target project
329
315
  - `run`: execute the decomposed wrapper, `work -> verify -> ship`, with optional apply
316
+ - `build`: one-command builder wrapper with `fast`, `safe`, `team`, `tdd`, and `release` modes
330
317
  - `report`: summarize session evidence into `REPORT.md` without project mutation
331
318
  - `review`: run the legacy full Claude-led/Codex-reviewed workflow
332
319
  - `review-cycle`: explicit compatibility alias for the legacy full review workflow
@@ -336,12 +323,15 @@ Advanced features such as `team-lite`, `ralph`, `wait`, instincts, cost tracking
336
323
 
337
324
  `plan` is recommended before `work` for larger changes. The current `run` command intentionally stays compact: it runs `work -> verify -> ship`, records acceptance criteria through `work`, and applies only when `--apply` is explicitly provided.
338
325
 
339
- Use `--profile quality` or `--profile security` on `work`, `verify`, and `run` when a task needs stronger evidence prompts. Add `--strict-quality` to `verify` or `run` when missing evidence or acceptance coverage should become a fix-required verdict before ship.
326
+ Use `build "<task>" --mode fast` when NEKOWORK should be the single entrypoint. Use `--mode safe` for security-sensitive work, `--mode team` for read-only multi-perspective planning before implementation, `--mode tdd` for strict acceptance evidence, and `--mode release` for readiness-focused work. `build` still uses one executor for writes, Codex verification before ship, and explicit apply only.
327
+
328
+ Use `--profile quality` or `--profile security` on `work`, `verify`, and `run` when a task needs stronger evidence prompts. Add `--strict-quality` to `verify`, `run`, or `build` when missing evidence or acceptance coverage should become a fix-required verdict before ship.
340
329
 
341
330
  Use official packs when choosing an install shape:
342
331
 
343
332
  ```bash
344
333
  node scripts/install-plan.js --list
334
+ node scripts/install-plan.js --pack builder
345
335
  node scripts/install-plan.js --pack quality
346
336
  node scripts/install-plan.js --pack security --target codex --json
347
337
  ```
@@ -354,8 +344,8 @@ Packs are aliases over validated profiles. They add clearer product packaging wi
354
344
  - Skills: 10
355
345
  - Hooks: 5
356
346
  - Modules: 7
357
- - Profiles: `core`, `developer`, `security`, `product`, `quality`, `frontend`, `testing`, `research`, `full`
358
- - Official packs: `core`, `quality`, `security`, `frontend`, `testing`, `release`, `enterprise`
347
+ - Profiles: `core`, `developer`, `builder`, `security`, `product`, `quality`, `frontend`, `testing`, `research`, `full`
348
+ - Official packs: `core`, `builder`, `quality`, `security`, `frontend`, `testing`, `release`, `enterprise`
359
349
  - Harness targets: `claude`, `codex`, `cursor`, `gemini`, `opencode`
360
350
 
361
351
  Key skills:
@@ -377,6 +367,8 @@ Key skills:
377
367
  node scripts/cli.js doctor
378
368
  node scripts/cli.js doctor --quick --gemini-smoke
379
369
  npm run demo:quick
370
+ node scripts/cli.js build "builder smoke" --mode team --session build-smoke
371
+ node scripts/cli.js report --session latest
380
372
  node scripts/install-plan.js --list
381
373
  node scripts/install-plan.js --pack quality
382
374
  node scripts/install-plan.js --profile developer
@@ -421,11 +413,12 @@ npm run security:hardening
421
413
  npm pack --dry-run --json
422
414
  ```
423
415
 
424
- `npm pack --dry-run --json` currently produces a package named like `ps-neko-nekowork-0.1.0-alpha.4.tgz`. It does not publish.
416
+ `npm pack --dry-run --json` currently produces a package named like `ps-neko-nekowork-0.1.0-alpha.6.tgz`. It does not publish.
425
417
 
426
418
  ## Documentation
427
419
 
428
420
  - [docs/QUICKSTART.md](docs/QUICKSTART.md) - first run and common paths
421
+ - [docs/BUILD.md](docs/BUILD.md) - build command modes and invariants
429
422
  - [docs/WHY-NEKOWORK.md](docs/WHY-NEKOWORK.md) - comparison and product positioning
430
423
  - [docs/CATALOG-PACKS.md](docs/CATALOG-PACKS.md) - curated catalog, official packs, and case-study evidence
431
424
  - [docs/PUBLISH-ALPHA.md](docs/PUBLISH-ALPHA.md) - public npm alpha release plan
@@ -446,7 +439,7 @@ npm pack --dry-run --json
446
439
  - [docs/RUNBOOK.md](docs/RUNBOOK.md) - operations guide
447
440
  - [docs/ARCHITECTURE.md](docs/ARCHITECTURE.md) - system architecture
448
441
  - [docs/PRODUCT-PRINCIPLES.md](docs/PRODUCT-PRINCIPLES.md) - product position, invariants, CLI phase semantics
449
- - [docs/AI-DEVELOPMENT-LIFECYCLE.md](docs/AI-DEVELOPMENT-LIFECYCLE.md) - quality runtime and disciplined AI development lifecycle
442
+ - [docs/AI-DEVELOPMENT-LIFECYCLE.md](docs/AI-DEVELOPMENT-LIFECYCLE.md) - AI development OS, quality runtime, and disciplined lifecycle
450
443
  - [docs/CORE-INVARIANTS.md](docs/CORE-INVARIANTS.md) - non-negotiable runtime safety rules
451
444
  - [docs/CLI-STAGES.md](docs/CLI-STAGES.md) - stage contract and compatibility transition
452
445
  - [docs/RISK-CLASSIFIER.md](docs/RISK-CLASSIFIER.md) - shared risk tags, challenge, and gate policy
package/agent.yaml CHANGED
@@ -1,8 +1,8 @@
1
1
  spec_version: gitagent/0.1.0
2
2
  name: nekowork
3
3
  runtime_name: harness
4
- version: 0.1.0-alpha.4
5
- description: "NEKOWORK HARNESS - Local-first multi-AI development verification runtime"
4
+ version: 0.1.0-alpha.6
5
+ description: "NEKOWORK HARNESS - Local-first AI development OS and quality runtime"
6
6
  license: MIT
7
7
  homepage: https://github.com/Ps-Neko/NEKOWORK
8
8
  # authors: contributor 목록은 git 히스토리로 갈음. 별도 명시 안 함.
@@ -89,6 +89,7 @@ profiles:
89
89
  available:
90
90
  - core
91
91
  - developer
92
+ - builder
92
93
  - security
93
94
  - product
94
95
  - quality
package/docs/ADVANCED.md CHANGED
@@ -1,6 +1,6 @@
1
1
  # Advanced Features
2
2
 
3
- The public alpha path focuses on `doctor`, `ask`, `plan`, `team`, `work`, `verify`, `gate`, `ship`, `apply`, `run`, `review`, `review-cycle`, and install/apply. This page keeps the larger runtime surface discoverable without crowding the first-run docs.
3
+ The public alpha path focuses on `doctor`, `build`, `report`, `gate`, and the decomposed `ask`, `plan`, `team`, `work`, `verify`, `ship`, `apply`, `run`, `review`, `review-cycle`, and install/apply surfaces. This page keeps the larger runtime surface discoverable without crowding the first-run docs.
4
4
 
5
5
  ## team
6
6
 
@@ -184,12 +184,37 @@ Outputs:
184
184
  - `.harness/state/sessions/<id>/run-summary.json`
185
185
  - all normal `work`, `verify`, `ship`, and optional `apply` outputs
186
186
 
187
+ ## build
188
+
189
+ `build` is the public one-command builder wrapper:
190
+
191
+ ```bash
192
+ node scripts/cli.js build "implement this safely" --mode fast --session build-smoke
193
+ node scripts/cli.js build "auth-sensitive change" --mode safe --session auth-smoke
194
+ node scripts/cli.js build "scope with team thinking" --mode team --session team-smoke
195
+ ```
196
+
197
+ Rules:
198
+
199
+ - `fast` runs the compact safe path through `run`.
200
+ - `safe` uses the security profile, strict quality, and Codex challenge.
201
+ - `team` creates read-only team handoffs before the single executor runs.
202
+ - `tdd` uses strict quality and acceptance evidence.
203
+ - `release` focuses on ship/readiness and report evidence.
204
+ - `apply` is never implicit; use `--apply` only for verified live-work diffs.
205
+
206
+ Outputs:
207
+
208
+ - `.harness/state/sessions/<id>/build-summary.json`
209
+ - all normal `run`, `work`, `verify`, `ship`, and optional `apply` outputs
210
+
187
211
  ## report
188
212
 
189
213
  `report` turns existing session evidence into a readable inspect-only report:
190
214
 
191
215
  ```bash
192
216
  node scripts/cli.js report --session run-smoke
217
+ node scripts/cli.js report --session latest
193
218
  node scripts/cli.js report --session run-smoke --stdout
194
219
  node scripts/cli.js report --session run-smoke --output docs/session-report.md
195
220
  ```
@@ -199,7 +224,7 @@ Rules:
199
224
  - Reads summaries, markers, acceptance criteria, and handoffs from `.harness/state/sessions/<id>/`.
200
225
  - Writes `REPORT.md` and `report-summary.json` by default.
201
226
  - Does not call providers, run git commands, apply diffs, or mutate target project files.
202
- - Can run after `ask`, `work`, `verify`, `ship`, `run`, or `apply`.
227
+ - Can run after `ask`, `work`, `verify`, `ship`, `run`, `build`, or `apply`.
203
228
 
204
229
  Outputs:
205
230
 
@@ -1,23 +1,24 @@
1
1
  # AI Development Lifecycle
2
2
 
3
- NEKOWORK is a local-first AI development quality runtime. Its job is not to collect every useful agent feature; its job is to make AI development disciplined, high-quality, independently verified, and human-gated.
3
+ NEKOWORK is a local-first AI development OS and quality runtime. Its job is not to collect every useful agent feature; its job is to make AI development fast, disciplined, high-quality, independently verified, and human-gated.
4
4
 
5
5
  ## Position
6
6
 
7
7
  ```text
8
- Good development habits
8
+ Fast build loop
9
+ + good development habits
9
10
  + quality rules, hooks, and skills
10
11
  + product-aware scope control
11
12
  + read-only multi-agent thinking
12
13
  + Codex verification
13
14
  + Human Gate and explicit apply
14
- = NEKOWORK gated AI development runtime
15
+ = NEKOWORK gated AI development OS
15
16
  ```
16
17
 
17
18
  The short slogan remains:
18
19
 
19
20
  ```text
20
- Claude work -> Codex verification -> Human Gate
21
+ fast AI build -> Codex verification -> Human Gate -> explicit apply
21
22
  ```
22
23
 
23
24
  ## Product Rules
@@ -29,6 +30,7 @@ Claude work -> Codex verification -> Human Gate
29
30
  5. Multiple agents may think, but only one executor writes.
30
31
  6. Rich skills, hooks, and rules may improve quality, but cannot weaken safety.
31
32
  7. Apply is explicit and evidence-based.
33
+ 8. Build modes may improve speed, but cannot bypass verification or gates.
32
34
 
33
35
  ## Absorption Model
34
36
 
@@ -40,6 +42,7 @@ External project ideas are absorbed as capabilities, not as a new architecture:
40
42
  | Rich agent environment | Skills, hooks, rules, MCP, memory, scanner-style checks | Profile/module based selective install |
41
43
  | Product questioning | Product, design, QA, release, and scope control questions | `ask`, `plan`, and `product` profile |
42
44
  | Team orchestration | Multiple perspectives and parallel review | `team` read-only handoffs |
45
+ | Productivity/autopilot-lite | One command for implementation, verification, and readiness | `build` modes over `run`; apply remains explicit |
43
46
  | NEKOWORK core | Codex verification, Human Gate, controlled apply | Non-bypassable runtime invariants |
44
47
 
45
48
  Capabilities can expand. The architecture cannot weaken the verification loop.
@@ -60,6 +63,16 @@ ask
60
63
 
61
64
  Quality enters early through `ask` and `plan`, not only at the final review step. Team mode collects multiple perspectives, but the write phase stays single-executor. Verification is independent, gate decisions are explicit, `report` makes evidence readable, and apply requires evidence.
62
65
 
66
+ For the one-command product experience, `build` selects a safe preset over the same loop:
67
+
68
+ ```bash
69
+ nekowork build "implement feature" --mode fast
70
+ nekowork build "auth-sensitive change" --mode safe
71
+ nekowork build "scope through QA first" --mode team
72
+ nekowork build "test-first change" --mode tdd
73
+ nekowork build "prepare release handoff" --mode release
74
+ ```
75
+
63
76
  ## Quality Profile
64
77
 
65
78
  The `quality` profile is the disciplined-development bundle:
@@ -1,6 +1,6 @@
1
1
  # Architecture
2
2
 
3
- NEKOWORK is the product. HARNESS is the local runtime packaged by NEKOWORK as a local-first AI development harness. The project keeps one canonical catalog and projects it into multiple agent surfaces.
3
+ NEKOWORK is the product. HARNESS is the local runtime packaged by NEKOWORK as a local-first AI development OS and quality harness. The project keeps one canonical catalog and projects it into multiple agent surfaces.
4
4
 
5
5
  ## Core Idea
6
6
 
@@ -25,10 +25,10 @@ The canonical source is the repository catalog. Generated harness directories ar
25
25
 
26
26
  ## Product Invariants
27
27
 
28
- NEKOWORK is a verification runtime, not a general agent pack:
28
+ NEKOWORK is an AI development OS bounded by a verification runtime, not a general agent pack:
29
29
 
30
30
  ```text
31
- Claude work -> Codex verification -> Human Gate
31
+ fast AI build -> Codex verification -> Human Gate -> explicit apply
32
32
  ```
33
33
 
34
34
  Core invariants:
@@ -50,7 +50,7 @@ User command
50
50
  |
51
51
  |-- doctor
52
52
  |-- install plan/apply
53
- |-- ask / plan / team / work / verify / gate / ship / report / apply / run / review / review-cycle
53
+ |-- ask / plan / team / work / verify / gate / ship / report / apply / run / build / review / review-cycle
54
54
  |-- ralph
55
55
  |-- team-lite
56
56
  |-- sessions / costs / instincts
@@ -87,6 +87,7 @@ node scripts/cli.js ship "ship readiness" --session work-smoke --project-root <t
87
87
  node scripts/cli.js report --session work-smoke --project-root <target>
88
88
  node scripts/cli.js apply --session work-smoke --project-root <target>
89
89
  node scripts/cli.js run "decomposed wrapper" --session run-smoke --project-root <target>
90
+ node scripts/cli.js build "safe builder wrapper" --mode team --session build-smoke --project-root <target>
90
91
  node scripts/cli.js review "change request" --no-ship --project-root <target>
91
92
  node scripts/cli.js review-cycle "legacy full-cycle request" --no-ship --project-root <target>
92
93
  ```
@@ -137,6 +138,8 @@ ask -> plan -> team -> work -> verify -> gate -> ship -> report -> apply
137
138
 
138
139
  `run` is the compatibility-friendly wrapper around the decomposed path. It runs `work -> verify -> ship` and only runs `apply` when `--apply` is explicitly requested and `SHIP_READY` exists. New automation should prefer `run` or the explicit decomposed commands; old automation can continue to use `review` or `review-cycle`.
139
140
 
141
+ `build` is the safe one-command builder entrypoint. It applies named mode presets (`fast`, `safe`, `team`, `tdd`, `release`) over the existing safe loop, records `build-summary.json`, and keeps apply explicit.
142
+
140
143
  `ralph` is an advanced repeated-iteration loop. Its default engine remains legacy `review` for compatibility, but `ralph --engine run` repeats the decomposed wrapper and records child run sessions. Ralph does not apply diffs; verified mutation still flows through `apply`.
141
144
 
142
145
  `wait` is an advanced persistent wakeup daemon. It watches session `wakeup.json` files, parses the session `active` contract, and resumes only supported modes (`ralph`, `run`, `review-cycle`). It writes `wait-summary.json` / `wait-events.jsonl`, backs off failed resumes, and refuses to resume sessions with `HUMAN_GATE`.
@@ -202,8 +205,8 @@ Builders project the catalog into tool-specific files:
202
205
 
203
206
  ## Release State
204
207
 
205
- The current release line is `0.1.0-alpha.4`:
208
+ The current repository release line is `0.1.0-alpha.6`:
206
209
 
207
210
  - Repository and GitHub tarball release are available.
208
- - Public npm alpha is published as `@ps-neko/nekowork@alpha`.
211
+ - Public npm alpha is published as `@ps-neko/nekowork@alpha` and currently points at `0.1.0-alpha.6`.
209
212
  - Clone, submodule, and local checkout integration remain supported for repository-pinned workflows.
package/docs/AUDIT.md CHANGED
@@ -2,22 +2,22 @@
2
2
 
3
3
  Status date: 2026-05-08
4
4
 
5
- This audit summarizes the current NEKOWORK state after publishing the `0.1.0-alpha.4` public alpha. It replaces the older week-by-week scratch audit, which contained stale planning notes and encoding damage.
5
+ This audit summarizes the current NEKOWORK state after publishing the `0.1.0-alpha.6` public alpha.
6
6
 
7
7
  ## Current Status
8
8
 
9
9
  | Area | Status | Notes |
10
10
  |---|---|---|
11
- | Package metadata | OK | `@ps-neko/nekowork@0.1.0-alpha.4`, `agent.yaml` uses `name: nekowork`, `runtime_name: harness`, and matching version |
12
- | npm publish | OK | `@ps-neko/nekowork@alpha` points at `0.1.0-alpha.4` |
11
+ | Package metadata | OK | repository version `@ps-neko/nekowork@0.1.0-alpha.6`, `agent.yaml` uses `name: nekowork`, `runtime_name: harness`, matching version, and `nekowork`/`harness` CLI bins |
12
+ | npm publish | OK | `@ps-neko/nekowork@alpha` points at `0.1.0-alpha.6` |
13
13
  | Source install | OK | Clone, local checkout, and submodule workflows are documented |
14
- | Public npm alpha | OK | `docs/PUBLISH-ALPHA.md` records the first alpha publish and the `0.1.0-alpha.4` alpha update |
14
+ | Public npm alpha | OK | `docs/PUBLISH-ALPHA.md` records alpha publishes through `0.1.0-alpha.6` |
15
15
  | CLI doctor/check | OK | `check`, `doctor`, `doctor --quick`, and `doctor --gemini-smoke` are available |
16
16
  | Provider auth | OK | Local delegated CLI auth is the default path |
17
17
  | Internal provider adapter | OK | `HARNESS_PROVIDER_OVERRIDE=internal` can call an explicit JSON command adapter without weakening gates |
18
- | Catalog | OK | 7 official packs, 11 agents, 10 skills, 5 hooks, 7 modules, 36 components, 9 profiles |
18
+ | Catalog | OK | 8 official packs, 11 agents, 10 skills, 5 hooks, 7 modules, 36 components, 10 profiles |
19
19
  | Multi-harness output | OK | Claude, Codex, Cursor, Gemini, and OpenCode builders are present |
20
- | Quick demo | OK | `npm run demo:quick` verifies the shortest no-API `doctor -> run -> report -> gate status` path |
20
+ | Quick demo | OK | `npm run demo:quick` verifies the shortest no-API `doctor -> build -> report -> gate status` path |
21
21
  | Fresh npm alpha smoke | OK | CI runs `npx -y @ps-neko/nekowork@alpha check --json` from a disposable directory |
22
22
  | Report UX | OK | `report` writes inspect-only `REPORT.md` and `report-summary.json` from session evidence |
23
23
  | External demo | OK | `npm run demo:external` verifies a disposable target project flow |
@@ -30,7 +30,7 @@ This audit summarizes the current NEKOWORK state after publishing the `0.1.0-alp
30
30
  | Persistent wakeup | OK | `wait` resumes supported active sessions and blocks on `HUMAN_GATE` |
31
31
  | Generated docs | OK | CODEMAP output is stable ASCII and reproducible |
32
32
  | Tests | OK | Unit, integration, and e2e suites pass locally and in CI |
33
- | Release | OK | `v0.1.0-alpha.4` is tagged and published as a GitHub prerelease |
33
+ | Release | OK | `v0.1.0-alpha.6` is tagged and published as a GitHub prerelease |
34
34
 
35
35
  ## Verification Gates
36
36
 
@@ -55,7 +55,7 @@ Current local result for this working tree:
55
55
  - `npm run test:unit`: covered by full `npm test`
56
56
  - `npm run validate:all`: pass
57
57
  - `npm run lint`: pass
58
- - `npm test`: 252 tests pass
58
+ - `npm test`: 263 tests pass
59
59
  - quick run demo: pass through `npm run demo:quick -- --cleanup`
60
60
  - external project e2e smoke: pass through `npm test`
61
61
  - `node scripts/sync-claude-md.js --check`: pass
@@ -63,13 +63,14 @@ Current local result for this working tree:
63
63
  - `npm audit --audit-level=moderate`: 0 vulnerabilities
64
64
  - `npm pack --dry-run --json`: pass
65
65
  - `npm publish --dry-run --access public --tag alpha`: pass
66
- - `npm publish --access public --tag alpha`: `0.1.0-alpha.4` published
67
- - `npm view @ps-neko/nekowork dist-tags version versions --json`: `alpha` points at `0.1.0-alpha.4`; `latest` remains `0.1.0-alpha.0`
68
- - `npx -y @ps-neko/nekowork@alpha check`: passed for `0.1.0-alpha.4` with WARN summary from non-git project root and Gemini auth not checked
66
+ - `npm publish --access public --tag alpha`: `0.1.0-alpha.6` published
67
+ - `npm view @ps-neko/nekowork dist-tags version versions --json`: `alpha` points at `0.1.0-alpha.6`; `latest` remains `0.1.0-alpha.0`
68
+ - `npx -y @ps-neko/nekowork@alpha check`: passed for `0.1.0-alpha.6` with WARN summary from Gemini auth not checked
69
69
 
70
70
  ## Completed Work
71
71
 
72
72
  - Local-first provider auth policy implemented and documented.
73
+ - `build` is the beginner entrypoint with `fast`, `safe`, `team`, `tdd`, and `release` modes over the safe run/report/gate loop.
73
74
  - Internal provider command adapter implemented and documented without bypassing verification, Human Gate, or apply controls.
74
75
  - `acceptance-coverage` skill added as a focused quality evidence helper.
75
76
  - API-key override warnings and guards are in place.
@@ -93,7 +94,7 @@ Current local result for this working tree:
93
94
  - Official packs expose curated install shapes without creating a second safety model.
94
95
  - Checked-in example fixtures now cover financial UI, CI hardening, and quality lifecycle evidence flows.
95
96
  - Third-party case studies record NEKOWORK runs against `sindresorhus/is-plain-obj`, `jshttp/basic-auth`, `python-hyper/h11`, and `motdotla/dotenv`.
96
- - Public npm alpha `0.1.0-alpha.4` is published under the `alpha` dist-tag.
97
+ - Public npm alpha `0.1.0-alpha.6` is published under the `alpha` dist-tag.
97
98
 
98
99
  ## Remaining Optional Work
99
100
 
@@ -113,7 +114,7 @@ Current local result for this working tree:
113
114
 
114
115
  ## External Readiness Score
115
116
 
116
- Current external readiness, excluding broader adoption evidence: **9.1 / 10**.
117
+ Current external readiness, excluding broader adoption evidence: **9.2 / 10**.
117
118
 
118
119
  Main deductions:
119
120