@jstn-sdk/ma 0.1.5 → 0.1.6
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/.agents/plugins/marketplace.json +20 -0
- package/.codex/hooks.json +1 -1
- package/.codex/prompts/enforcement.md +3 -3
- package/.codex/prompts/onboarding.md +29 -14
- package/README.md +512 -129
- package/bin/ma.js +61 -103
- package/docs/README.md +2 -1
- package/docs/assets/image/Screenshot(1).png +0 -0
- package/docs/assets/image/Screenshot(2).png +0 -0
- package/docs/assets/image/Screenshot(3).png +0 -0
- package/docs/assets/image/Screenshot(4).png +0 -0
- package/docs/assets/image/Screenshot(5).png +0 -0
- package/docs/assets/image/Screenshot(6).png +0 -0
- package/docs/assets/image/Screenshot(7).png +0 -0
- package/docs/assets/image/Screenshot(8).png +0 -0
- package/docs/assets/image/Screenshot(9).png +0 -0
- package/docs/assets/meta-architect-logo.png +0 -0
- package/docs/assets/meta-architect-logo.svg +8 -0
- package/docs/getting-started.md +419 -37
- package/docs/installed-sdk.md +60 -0
- package/docs/mcp-setup.md +3 -134
- package/docs/onboarding.md +41 -11
- package/docs/qa/release-readiness-0.1.5.md +46 -44
- package/docs/qa/release-readiness-0.1.6.md +79 -0
- package/docs/release-spec.md +81 -103
- package/docs/skills-publishing.md +187 -54
- package/docs/skills.md +82 -141
- package/mcp/collections.json +1 -1
- package/mcp/fallback.json +1 -1
- package/mcp/servers.json +1 -1
- package/package.json +17 -14
- package/plugins/meta-architect/.app.json +1 -1
- package/plugins/meta-architect/.codex-plugin/plugin.json +23 -0
- package/plugins/meta-architect/.mcp.json +1 -1
- package/plugins/meta-architect/README.md +23 -10
- package/plugins/meta-architect/skills/arch/SKILL.md +27 -0
- package/plugins/meta-architect/skills/arch/agents/openai.yaml +4 -0
- package/plugins/meta-architect/skills/build/SKILL.md +24 -0
- package/plugins/meta-architect/skills/build/agents/openai.yaml +4 -0
- package/plugins/meta-architect/skills/flow/SKILL.md +24 -0
- package/plugins/meta-architect/skills/flow/agents/openai.yaml +4 -0
- package/plugins/meta-architect/skills/maestro/SKILL.md +25 -0
- package/plugins/meta-architect/skills/maestro/agents/openai.yaml +4 -0
- package/plugins/meta-architect/skills/meta-architect/SKILL.md +17 -23
- package/plugins/meta-architect/skills/meta-architect/agents/openai.yaml +4 -0
- package/plugins/meta-architect/skills/meta-architect/references/core-release-rules.md +13 -0
- package/plugins/meta-architect/skills/sage/SKILL.md +24 -0
- package/plugins/meta-architect/skills/sage/agents/openai.yaml +4 -0
- package/plugins/meta-architect/skills/vet/SKILL.md +25 -0
- package/plugins/meta-architect/skills/vet/agents/openai.yaml +4 -0
- package/plugins/meta-architect/skills/vibe/SKILL.md +24 -0
- package/plugins/meta-architect/skills/vibe/agents/openai.yaml +4 -0
- package/scripts/doctor.js +8 -0
- package/scripts/plugin-sync.js +93 -0
- package/scripts/postinstall.js +23 -0
- package/scripts/release-metadata.js +94 -0
- package/scripts/release-sync.js +359 -0
- package/scripts/release-verify.js +153 -0
- package/scripts/setup-npmrc.js +39 -0
- package/scripts/skills-install.js +4 -36
- package/scripts/skills-manifest.js +1 -1
- package/scripts/skills-validate.js +78 -40
- package/skills/arch/SKILL.md +27 -0
- package/skills/arch/agents/openai.yaml +4 -0
- package/skills/build/SKILL.md +24 -0
- package/skills/build/agents/openai.yaml +4 -0
- package/skills/flow/SKILL.md +24 -0
- package/skills/flow/agents/openai.yaml +4 -0
- package/skills/index.json +27 -22
- package/skills/maestro/SKILL.md +25 -0
- package/skills/maestro/agents/openai.yaml +4 -0
- package/skills/meta-architect/SKILL.md +17 -23
- package/skills/meta-architect/agents/openai.yaml +3 -3
- package/skills/meta-architect/references/core-release-rules.md +2 -2
- package/skills/sage/SKILL.md +24 -0
- package/skills/sage/agents/openai.yaml +4 -0
- package/skills/vet/SKILL.md +25 -0
- package/skills/vet/agents/openai.yaml +4 -0
- package/skills/vibe/SKILL.md +24 -0
- package/skills/vibe/agents/openai.yaml +4 -0
- package/sprint/00-idea.md +1 -1
- package/sprint/01-architecture.md +1 -1
- package/sprint/02-oss-evidence.md +1 -1
- package/sprint/03-logic.md +1 -1
- package/sprint/04-security.md +2 -2
- package/sprint/05-dx-ux.md +1 -1
- package/src/decision-log.js +4 -4
- package/src/launcher.js +21 -17
- package/src/mcp-live-client.js +1 -1
- package/src/paths.js +8 -32
- package/src/release-state.js +3 -3
- package/src/runtime-artifacts.js +411 -0
- package/src/skill-installer.js +198 -0
- package/src/skills.js +473 -166
- package/templates/AGENTS.md +6 -6
- package/templates/model-instructions/core.md +1 -1
- package/.codex/config.toml +0 -2
- package/plugins/meta-architect/skills/meta-architect-arch/SKILL.md +0 -24
- package/plugins/meta-architect/skills/meta-architect-build/SKILL.md +0 -25
- package/plugins/meta-architect/skills/meta-architect-flow/SKILL.md +0 -23
- package/plugins/meta-architect/skills/meta-architect-sage/SKILL.md +0 -23
- package/plugins/meta-architect/skills/meta-architect-vet/SKILL.md +0 -23
- package/plugins/meta-architect/skills/meta-architect-vibe/SKILL.md +0 -24
- package/prompts/architect.md +0 -216
- package/prompts/builder.md +0 -10
- package/prompts/flow.md +0 -9
- package/prompts/release-manager.md +0 -10
- package/prompts/sage.md +0 -10
- package/prompts/security-reviewer.md +0 -10
- package/prompts/verifier.md +0 -10
- package/prompts/vibe.md +0 -10
- package/skills/meta-architect-arch/SKILL.md +0 -24
- package/skills/meta-architect-arch/agents/openai.yaml +0 -4
- package/skills/meta-architect-build/SKILL.md +0 -25
- package/skills/meta-architect-build/agents/openai.yaml +0 -4
- package/skills/meta-architect-flow/SKILL.md +0 -23
- package/skills/meta-architect-flow/agents/openai.yaml +0 -4
- package/skills/meta-architect-sage/SKILL.md +0 -23
- package/skills/meta-architect-sage/agents/openai.yaml +0 -4
- package/skills/meta-architect-vet/SKILL.md +0 -23
- package/skills/meta-architect-vet/agents/openai.yaml +0 -4
- package/skills/meta-architect-vibe/SKILL.md +0 -24
- package/skills/meta-architect-vibe/agents/openai.yaml +0 -4
- package/src/doctor.js +0 -30
- package/src/setup.js +0 -375
package/docs/getting-started.md
CHANGED
|
@@ -1,24 +1,169 @@
|
|
|
1
1
|
# Getting Started
|
|
2
2
|
|
|
3
|
-
|
|
3
|
+
This guide takes Meta-Architect from install to a green-gated skill workflow on a real project.
|
|
4
4
|
|
|
5
|
-
##
|
|
5
|
+
## Goal
|
|
6
6
|
|
|
7
|
-
|
|
7
|
+
By the end of this guide you should be able to:
|
|
8
|
+
- install Meta-Architect
|
|
9
|
+
- configure MCP/GitMCP sources
|
|
10
|
+
- run the full architecture-to-build sequence
|
|
11
|
+
- understand how each gate moves
|
|
12
|
+
- diagnose blocked gates
|
|
13
|
+
- use the merge/release path safely
|
|
14
|
+
|
|
15
|
+
## Prerequisites
|
|
16
|
+
|
|
17
|
+
- Node.js 20+
|
|
18
|
+
- npm 10+
|
|
19
|
+
- Git 2.30+
|
|
20
|
+
- an MCP-capable runtime
|
|
21
|
+
- network access if you want live `$sage` verification against a real GitMCP server
|
|
22
|
+
|
|
23
|
+
## 1. Canonical install and launch
|
|
24
|
+
|
|
25
|
+
Canonical public install:
|
|
8
26
|
|
|
9
27
|
```bash
|
|
10
|
-
|
|
28
|
+
# Install
|
|
29
|
+
npm i -g @openai/codex@latest @jstn-sdk/ma@latest
|
|
30
|
+
|
|
31
|
+
# Launch
|
|
32
|
+
ma --madmax --high
|
|
33
|
+
|
|
34
|
+
# Remove Meta-Architect only
|
|
35
|
+
npm uninstall -g @jstn-sdk/ma
|
|
36
|
+
|
|
37
|
+
# Remove Meta-Architect and Codex
|
|
38
|
+
npm uninstall -g @jstn-sdk/ma @openai/codex
|
|
11
39
|
```
|
|
12
40
|
|
|
13
|
-
|
|
41
|
+
That is the default product path. The product experience is the in-session skill workflow in [example/usage-workflow.md](../example/usage-workflow.md). The `ma` command is only a helper for starting or supporting that flow.
|
|
42
|
+
|
|
43
|
+
## 2. Real usage workflow
|
|
44
|
+
|
|
45
|
+
Start with the structured `$arch` prompt:
|
|
46
|
+
|
|
47
|
+
```text
|
|
48
|
+
$arch I want to build: [PROJECT IDEA]
|
|
49
|
+
|
|
50
|
+
Context:
|
|
51
|
+
- Product type: [web app / mobile app / API / marketplace / agent system / internal tool]
|
|
52
|
+
- Users: [who will use it]
|
|
53
|
+
- Core problem: [what problem it solves]
|
|
54
|
+
- Main features:
|
|
55
|
+
1. [feature one]
|
|
56
|
+
2. [feature two]
|
|
57
|
+
3. [feature three]
|
|
58
|
+
- Constraints:
|
|
59
|
+
- Budget: [low / medium / high]
|
|
60
|
+
- Team size: [solo / small / medium]
|
|
61
|
+
- Timeline: [e.g. 2 weeks MVP, 3 months beta]
|
|
62
|
+
- Preferred stack: [optional]
|
|
63
|
+
- Avoid: [optional]
|
|
64
|
+
- Quality priorities:
|
|
65
|
+
- [e.g. speed, low cost, security, DX, maintainability, scalability]
|
|
66
|
+
- Deployment target:
|
|
67
|
+
- [Vercel / Docker / VPS / AWS / GCP / local-first / hybrid]
|
|
68
|
+
|
|
69
|
+
Required output:
|
|
70
|
+
1. Problem framing
|
|
71
|
+
2. Recommended architecture
|
|
72
|
+
3. Stack decision with justification
|
|
73
|
+
4. System components and responsibilities
|
|
74
|
+
5. Data model and storage choices
|
|
75
|
+
6. Auth/security considerations
|
|
76
|
+
7. DX/UX considerations
|
|
77
|
+
8. Delivery plan for v0.1.6
|
|
78
|
+
9. Risks and trade-offs
|
|
79
|
+
10. Decision log
|
|
80
|
+
11. Exact next trigger to run after this
|
|
81
|
+
```
|
|
82
|
+
|
|
83
|
+
Then continue with:
|
|
84
|
+
|
|
85
|
+
```text
|
|
86
|
+
$sage
|
|
87
|
+
$flow
|
|
88
|
+
$vet
|
|
89
|
+
$vibe
|
|
90
|
+
$build
|
|
91
|
+
```
|
|
92
|
+
|
|
93
|
+
Use the full prompt blocks from [example/usage-workflow.md](../example/usage-workflow.md) when you want the exact handoff format between lanes.
|
|
94
|
+
|
|
95
|
+
## 3. Contributor clone and link
|
|
14
96
|
|
|
15
97
|
```bash
|
|
16
|
-
|
|
98
|
+
git clone https://github.com/JustineDevs/meta-architect.git
|
|
99
|
+
cd meta-architect
|
|
100
|
+
npm install
|
|
101
|
+
npm link
|
|
102
|
+
```
|
|
103
|
+
|
|
104
|
+
Why `npm link` matters:
|
|
105
|
+
- it makes the local helper command available without requiring a global publish step
|
|
106
|
+
|
|
107
|
+
## 4. Secondary repository setup flow
|
|
108
|
+
|
|
109
|
+
```bash
|
|
110
|
+
ma setup
|
|
111
|
+
```
|
|
112
|
+
|
|
113
|
+
Optional interactive session:
|
|
114
|
+
|
|
115
|
+
```bash
|
|
116
|
+
ma
|
|
17
117
|
```
|
|
18
118
|
|
|
19
|
-
|
|
119
|
+
Expected effects:
|
|
120
|
+
- `.codex/agents/` exists
|
|
121
|
+
- `.codex/prompts/` exists
|
|
122
|
+
- local `.ma/skills/`, `.ma/evidence/`, `.ma/context/`, `.ma/specs/`, `.ma/plans/`, and `.ma/runbook.md` are seeded
|
|
123
|
+
- `mcp/`, `docs/`, and `sprint/` surfaces exist
|
|
124
|
+
- `ma` opens Codex with the Meta-Architect helper posture when run with no arguments
|
|
125
|
+
|
|
126
|
+
Expected output:
|
|
127
|
+
|
|
128
|
+
```text
|
|
129
|
+
meta-architect setup
|
|
130
|
+
====================
|
|
131
|
+
ready: .codex/agents
|
|
132
|
+
ready: .codex/prompts
|
|
133
|
+
ready: .ma/skills
|
|
134
|
+
ready: .ma/evidence
|
|
135
|
+
ready: .ma/context
|
|
136
|
+
ready: .ma/specs
|
|
137
|
+
ready: .ma/plans
|
|
138
|
+
ready: mcp
|
|
139
|
+
ready: docs
|
|
140
|
+
ready: docs/qa
|
|
141
|
+
ready: sprint
|
|
142
|
+
```
|
|
143
|
+
|
|
144
|
+
This setup path is secondary to the canonical in-session skill workflow. Use it when you need local repo scaffolding.
|
|
145
|
+
|
|
146
|
+
## 4.1 Understand the two surfaces
|
|
147
|
+
|
|
148
|
+
Meta-Architect works in two simple ways:
|
|
149
|
+
|
|
150
|
+
- terminal commands
|
|
151
|
+
- in-session skills
|
|
152
|
+
|
|
153
|
+
Terminal commands are normal shell commands:
|
|
154
|
+
|
|
155
|
+
```bash
|
|
156
|
+
ma setup
|
|
157
|
+
ma init
|
|
158
|
+
ma sdk-path
|
|
159
|
+
ma status
|
|
160
|
+
ma run '$arch'
|
|
161
|
+
```
|
|
162
|
+
|
|
163
|
+
In-session skills are prompts used after you are already inside Codex:
|
|
20
164
|
|
|
21
165
|
```text
|
|
166
|
+
$maestro
|
|
22
167
|
$arch
|
|
23
168
|
$sage
|
|
24
169
|
$flow
|
|
@@ -27,62 +172,299 @@ $vibe
|
|
|
27
172
|
$build
|
|
28
173
|
```
|
|
29
174
|
|
|
30
|
-
|
|
175
|
+
Easy rule:
|
|
176
|
+
- `ma ...` = terminal helper command
|
|
177
|
+
- `$...` = in-session skill
|
|
178
|
+
|
|
179
|
+
`ma setup` and `ma init` currently do the same thing:
|
|
180
|
+
- they create local `.ma/` support files
|
|
181
|
+
- they prepare context, specs, plans, evidence, and runbook files
|
|
182
|
+
- they do not automatically run the skill workflow
|
|
183
|
+
|
|
184
|
+
`ma sdk-path` prints the installed packaged support-bundle root for relevant files such as prompts, MCP files, sprint files, scripts, plugin metadata, and templates.
|
|
185
|
+
|
|
186
|
+
## 5. Configure MCP / GitMCP
|
|
187
|
+
|
|
188
|
+
Edit:
|
|
189
|
+
- `mcp/servers.json`
|
|
190
|
+
- `mcp/collections.json`
|
|
191
|
+
- `mcp/fallback.json`
|
|
192
|
+
|
|
193
|
+
Minimum live example:
|
|
194
|
+
|
|
195
|
+
```json
|
|
196
|
+
{
|
|
197
|
+
"category": "meta-list",
|
|
198
|
+
"repo": "sindresorhus/awesome",
|
|
199
|
+
"endpoint": "https://gitmcp.io/sindresorhus/awesome"
|
|
200
|
+
}
|
|
201
|
+
```
|
|
202
|
+
|
|
203
|
+
Recommended first set:
|
|
204
|
+
- `sindresorhus/awesome`
|
|
205
|
+
- `dzharii/awesome-typescript`
|
|
206
|
+
- `sbilly/awesome-security`
|
|
207
|
+
|
|
208
|
+
See [docs/mcp-setup.md](./mcp-setup.md) for endpoint policy and evidence semantics.
|
|
209
|
+
|
|
210
|
+
## 6. Secondary helper flow
|
|
31
211
|
|
|
32
|
-
|
|
33
|
-
|
|
34
|
-
|
|
35
|
-
- `ma` is a runtime launcher plus a small maintenance/helper interface.
|
|
212
|
+
```bash
|
|
213
|
+
ma idea "Build a real-time collaborative whiteboard for product teams"
|
|
214
|
+
```
|
|
36
215
|
|
|
37
|
-
|
|
216
|
+
Expected effects:
|
|
217
|
+
- an idea decision entry is appended to `.ma/decisions.json`
|
|
218
|
+
- `idea_status = CLEAR`
|
|
38
219
|
|
|
39
|
-
|
|
220
|
+
If this fails:
|
|
221
|
+
- ensure the idea text is not empty
|
|
222
|
+
- inspect `.ma/decisions.json` for malformed local data
|
|
223
|
+
|
|
224
|
+
## 7. Run the helper skill sequence
|
|
225
|
+
|
|
226
|
+
### 5.0 Workflow manager
|
|
227
|
+
|
|
228
|
+
```bash
|
|
229
|
+
ma run '$maestro'
|
|
230
|
+
```
|
|
231
|
+
|
|
232
|
+
Expected effects:
|
|
233
|
+
- reads the current gate state
|
|
234
|
+
- recommends the best next step
|
|
235
|
+
- writes `.ma/plans/maestro.md`
|
|
236
|
+
- records an advisory decision entry
|
|
237
|
+
|
|
238
|
+
Generated or updated:
|
|
239
|
+
- `.ma/decisions.json`
|
|
240
|
+
- `.ma/plans/maestro.md`
|
|
241
|
+
|
|
242
|
+
### 5.1 Architecture
|
|
40
243
|
|
|
41
244
|
```bash
|
|
42
|
-
ma status
|
|
43
|
-
ma idea "Build a real-time collaborative whiteboard"
|
|
44
245
|
ma run '$arch'
|
|
246
|
+
```
|
|
247
|
+
|
|
248
|
+
Expected effects:
|
|
249
|
+
- a structured first-pass architecture blueprint is appended
|
|
250
|
+
- `architecture_status = APPROVED`
|
|
251
|
+
|
|
252
|
+
Generated or updated:
|
|
253
|
+
- `.ma/decisions.json`
|
|
254
|
+
- `.ma/release.json`
|
|
255
|
+
- `.ma/context/project.md`
|
|
256
|
+
- `.ma/specs/architecture.md`
|
|
257
|
+
- `.ma/plans/implementation.md`
|
|
258
|
+
|
|
259
|
+
### 5.2 Evidence
|
|
260
|
+
|
|
261
|
+
```bash
|
|
45
262
|
ma run '$sage'
|
|
263
|
+
```
|
|
264
|
+
|
|
265
|
+
Expected effects:
|
|
266
|
+
- `architecture_status` must already be approved
|
|
267
|
+
- configured GitMCP endpoints are validated
|
|
268
|
+
- the latest architecture summary is used as the probe query basis
|
|
269
|
+
- the first configured live source is probed when live mode is enabled
|
|
270
|
+
- `.ma/evidence/sources.json` is updated
|
|
271
|
+
- `evidence_status` becomes:
|
|
272
|
+
- `VERIFIED` on a real successful live probe
|
|
273
|
+
- `PARTIAL` when configured evidence exists but live proof is incomplete or disabled
|
|
274
|
+
- `MISSING` when no usable approved source exists
|
|
275
|
+
|
|
276
|
+
Generated or updated:
|
|
277
|
+
- `.ma/evidence/sources.json`
|
|
278
|
+
- `.ma/decisions.json`
|
|
279
|
+
- `.ma/release.json`
|
|
280
|
+
- `.ma/specs/evidence.md`
|
|
281
|
+
|
|
282
|
+
If this fails:
|
|
283
|
+
- check endpoint URLs in `mcp/servers.json`
|
|
284
|
+
- ensure architecture was approved first
|
|
285
|
+
- verify network access
|
|
286
|
+
- rerun after correcting the endpoint or environment
|
|
287
|
+
|
|
288
|
+
### 5.3 Logic
|
|
289
|
+
|
|
290
|
+
```bash
|
|
46
291
|
ma run '$flow'
|
|
292
|
+
```
|
|
293
|
+
|
|
294
|
+
Expected effects:
|
|
295
|
+
- a structured first-pass logic/state review is appended
|
|
296
|
+
- `logic_status = GREEN` when current prerequisites and transition modeling are acceptable
|
|
297
|
+
- `logic_status = RED` when prerequisite gates are not ready
|
|
298
|
+
|
|
299
|
+
Generated or updated:
|
|
300
|
+
- `.ma/specs/logic.md`
|
|
301
|
+
|
|
302
|
+
### 5.4 Security
|
|
303
|
+
|
|
304
|
+
```bash
|
|
47
305
|
ma run '$vet'
|
|
306
|
+
```
|
|
307
|
+
|
|
308
|
+
Expected effects:
|
|
309
|
+
- `.ma/evidence/audits.json` and `.ma/evidence/cves.json` are updated
|
|
310
|
+
- `security_status = GREEN` on a baseline pass
|
|
311
|
+
- `security_status = RED` when prerequisite gates are not ready
|
|
312
|
+
|
|
313
|
+
Generated or updated:
|
|
314
|
+
- `.ma/specs/security.md`
|
|
315
|
+
|
|
316
|
+
### 5.5 Experience
|
|
317
|
+
|
|
318
|
+
```bash
|
|
48
319
|
ma run '$vibe'
|
|
320
|
+
ma run '$vibe' --waive --reason "Accepted for this release line"
|
|
321
|
+
```
|
|
322
|
+
|
|
323
|
+
Expected effects:
|
|
324
|
+
- `.ma/evidence/outcomes.json` is updated
|
|
325
|
+
- `experience_status = GREEN` on a baseline pass
|
|
326
|
+
- `experience_status = RED` when prerequisite gates are not ready
|
|
327
|
+
- `experience_status = WAIVED` when the waiver path is used explicitly
|
|
328
|
+
|
|
329
|
+
Generated or updated:
|
|
330
|
+
- `.ma/specs/experience.md`
|
|
331
|
+
|
|
332
|
+
## 8. Inspect gate status
|
|
333
|
+
|
|
334
|
+
```bash
|
|
335
|
+
ma status
|
|
336
|
+
```
|
|
337
|
+
|
|
338
|
+
Expected green-state output:
|
|
339
|
+
|
|
340
|
+
```text
|
|
341
|
+
Meta-Architect Status
|
|
342
|
+
=====================
|
|
343
|
+
Idea: CLEAR
|
|
344
|
+
Architecture: APPROVED
|
|
345
|
+
Evidence: VERIFIED
|
|
346
|
+
Logic: GREEN
|
|
347
|
+
Security: GREEN
|
|
348
|
+
Experience: GREEN
|
|
349
|
+
Build: LOCKED
|
|
350
|
+
Next allowed triggers:
|
|
351
|
+
$build
|
|
352
|
+
```
|
|
353
|
+
|
|
354
|
+
## 9. Unlock and run build planning
|
|
355
|
+
|
|
356
|
+
```bash
|
|
49
357
|
ma run '$build'
|
|
50
|
-
|
|
358
|
+
```
|
|
359
|
+
|
|
360
|
+
Expected effects:
|
|
361
|
+
- build gate is evaluated
|
|
362
|
+
- if allowed, `build_status = READY`
|
|
363
|
+
- branch suggestions are printed
|
|
364
|
+
- worktree commands are suggested
|
|
365
|
+
- `.ma/plans/build.md` is updated
|
|
366
|
+
|
|
367
|
+
Expected output shape:
|
|
368
|
+
|
|
369
|
+
```text
|
|
370
|
+
Build gate is green.
|
|
371
|
+
Suggested branches:
|
|
372
|
+
- feature/ui
|
|
373
|
+
- feature/api
|
|
374
|
+
Optional worktree commands:
|
|
375
|
+
git worktree add ../ui feature/ui
|
|
376
|
+
git worktree add ../api feature/api
|
|
377
|
+
```
|
|
378
|
+
|
|
379
|
+
If `$build` fails:
|
|
380
|
+
- run `ma status`
|
|
381
|
+
- read the blocking statuses
|
|
382
|
+
- fix the corresponding upstream lane
|
|
383
|
+
- rerun that lane, then rerun `$build`
|
|
384
|
+
|
|
385
|
+
## 10. Example walkthrough: collaborative whiteboard
|
|
386
|
+
|
|
387
|
+
```bash
|
|
51
388
|
ma setup
|
|
389
|
+
ma idea "Build a collaborative whiteboard with live cursors and shared boards"
|
|
390
|
+
ma run '$arch'
|
|
391
|
+
ma run '$sage'
|
|
392
|
+
ma run '$flow'
|
|
393
|
+
ma run '$vet'
|
|
394
|
+
ma run '$vibe'
|
|
395
|
+
ma status
|
|
396
|
+
ma run '$build'
|
|
52
397
|
```
|
|
53
398
|
|
|
54
|
-
|
|
55
|
-
- `AGENTS.md`
|
|
56
|
-
- `.codex/`
|
|
57
|
-
- `.omx/`
|
|
58
|
-
- `mcp/`
|
|
59
|
-
- `templates/`
|
|
399
|
+
If you want an interactive Codex session during the walkthrough, start it separately with `ma`.
|
|
60
400
|
|
|
61
|
-
|
|
401
|
+
What should happen:
|
|
402
|
+
- `$arch` records a structured first-pass blueprint
|
|
403
|
+
- `$sage` binds major choices to configured GitMCP-backed sources
|
|
404
|
+
- `$flow` records the kernel’s baseline state review for the mission
|
|
405
|
+
- `$vet` records a baseline security review
|
|
406
|
+
- `$vibe` records baseline DX/UX guidance
|
|
407
|
+
- `$build` suggests bounded concerns like `feature/ui` and `feature/api`
|
|
62
408
|
|
|
63
|
-
|
|
409
|
+
Related mission:
|
|
410
|
+
- [missions/collaborative-whiteboard/mission.md](../missions/collaborative-whiteboard/mission.md)
|
|
64
411
|
|
|
65
|
-
|
|
412
|
+
## 9. Merge and release path
|
|
66
413
|
|
|
67
|
-
|
|
68
|
-
|
|
69
|
-
|
|
70
|
-
|
|
414
|
+
After implementation work is complete:
|
|
415
|
+
|
|
416
|
+
```bash
|
|
417
|
+
ma merge feature/ui development
|
|
418
|
+
ma release development prod
|
|
419
|
+
```
|
|
420
|
+
|
|
421
|
+
Expected effects:
|
|
422
|
+
- merge only succeeds for `feature/* -> development`
|
|
423
|
+
- release only succeeds for `development|release/* -> prod`
|
|
424
|
+
- final statuses advance to:
|
|
425
|
+
- `build_status = DONE`
|
|
426
|
+
- `merge_status = MERGED_TO_DEVELOPMENT`
|
|
427
|
+
- `release_status = SHIPPED_TO_PROD`
|
|
428
|
+
|
|
429
|
+
## 10. Files generated or updated during a normal run
|
|
430
|
+
|
|
431
|
+
- `.ma/decisions.json`
|
|
432
|
+
- `.ma/release.json`
|
|
433
|
+
- `.ma/evidence/sources.json`
|
|
434
|
+
- `.ma/evidence/audits.json`
|
|
435
|
+
- `.ma/evidence/cves.json`
|
|
436
|
+
- `.ma/evidence/outcomes.json`
|
|
437
|
+
- `.ma/context/project.md`
|
|
438
|
+
- `.ma/specs/architecture.md`
|
|
439
|
+
- `.ma/specs/evidence.md`
|
|
440
|
+
- `.ma/specs/logic.md`
|
|
441
|
+
- `.ma/specs/security.md`
|
|
442
|
+
- `.ma/specs/experience.md`
|
|
443
|
+
- `.ma/plans/implementation.md`
|
|
444
|
+
- `.ma/plans/build.md`
|
|
445
|
+
- `.ma/runbook.md`
|
|
71
446
|
|
|
72
|
-
|
|
447
|
+
These are local product artifacts created by the runtime. They are not a reason to bypass gate logic manually.
|
|
73
448
|
|
|
74
|
-
##
|
|
449
|
+
## 11. If a gate fails
|
|
75
450
|
|
|
76
|
-
|
|
451
|
+
Rule:
|
|
452
|
+
- do not edit statuses manually
|
|
453
|
+
- rerun the correct upstream command
|
|
454
|
+
- use `ma status` as the authority
|
|
77
455
|
|
|
78
|
-
|
|
79
|
-
- `
|
|
80
|
-
-
|
|
81
|
-
-
|
|
456
|
+
Common examples:
|
|
457
|
+
- `evidence_status = MISSING` -> fix `mcp/servers.json`, rerun `$sage`
|
|
458
|
+
- `logic_status = RED` -> fix upstream evidence/architecture issues, rerun `$flow`
|
|
459
|
+
- `security_status = RED` -> resolve issues surfaced by `$vet`
|
|
460
|
+
- `experience_status = RED` -> rerun `$vibe` after clearing prerequisites
|
|
461
|
+
- `experience_status = WAIVED` -> verify the waiver reason recorded in `.ma/release.json`
|
|
82
462
|
|
|
83
463
|
## Related docs
|
|
84
464
|
|
|
465
|
+
- [README.md](../README.md)
|
|
85
466
|
- [Skills Reference](./skills.md)
|
|
86
467
|
- [MCP Setup](./mcp-setup.md)
|
|
87
468
|
- [Release Spec](./release-spec.md)
|
|
469
|
+
- [Skills Publishing](./skills-publishing.md)
|
|
88
470
|
- [Plugin Bundle](../plugins/meta-architect/README.md)
|
|
@@ -0,0 +1,60 @@
|
|
|
1
|
+
# Installed Support Bundle
|
|
2
|
+
|
|
3
|
+
Meta-Architect installs two things into the active Codex home:
|
|
4
|
+
|
|
5
|
+
1. skills under `CODEX_HOME/skills/`
|
|
6
|
+
2. a support bundle under `CODEX_HOME/meta-architect-sdk/`
|
|
7
|
+
|
|
8
|
+
The support bundle exists so skills and helper paths can use relevant packaged files without guessing where they live.
|
|
9
|
+
|
|
10
|
+
## Canonical support bundle path
|
|
11
|
+
|
|
12
|
+
Default:
|
|
13
|
+
|
|
14
|
+
```text
|
|
15
|
+
~/.codex/meta-architect-sdk/
|
|
16
|
+
```
|
|
17
|
+
|
|
18
|
+
If `CODEX_HOME` is set, use:
|
|
19
|
+
|
|
20
|
+
```text
|
|
21
|
+
$CODEX_HOME/meta-architect-sdk/
|
|
22
|
+
```
|
|
23
|
+
|
|
24
|
+
You can print the exact active path with:
|
|
25
|
+
|
|
26
|
+
```bash
|
|
27
|
+
ma sdk-path
|
|
28
|
+
```
|
|
29
|
+
|
|
30
|
+
## What is installed there
|
|
31
|
+
|
|
32
|
+
- `mcp/`
|
|
33
|
+
- `sprint/`
|
|
34
|
+
- `prompts/`
|
|
35
|
+
- `scripts/`
|
|
36
|
+
- `plugins/meta-architect/`
|
|
37
|
+
- `templates/`
|
|
38
|
+
- `docs/README.md`
|
|
39
|
+
- `asset-manifest.json`
|
|
40
|
+
|
|
41
|
+
## Why this exists
|
|
42
|
+
|
|
43
|
+
- skills can rely on a standard installed asset root
|
|
44
|
+
- helper paths can repair missing support assets automatically
|
|
45
|
+
- packaged references do not depend on a source checkout
|
|
46
|
+
- the product can use relevant packaged files without path guessing
|
|
47
|
+
|
|
48
|
+
## Contract
|
|
49
|
+
|
|
50
|
+
When Meta-Architect needs packaged support files, prefer the installed support bundle first.
|
|
51
|
+
|
|
52
|
+
Examples:
|
|
53
|
+
- MCP starter files -> `meta-architect-sdk/mcp/`
|
|
54
|
+
- sprint references -> `meta-architect-sdk/sprint/`
|
|
55
|
+
- Codex prompt assets -> `meta-architect-sdk/prompts/`
|
|
56
|
+
- helper scripts -> `meta-architect-sdk/scripts/`
|
|
57
|
+
- plugin metadata -> `meta-architect-sdk/plugins/meta-architect/`
|
|
58
|
+
- templates -> `meta-architect-sdk/templates/`
|
|
59
|
+
|
|
60
|
+
The in-session skill flow is still primary. This bundle only standardizes packaged asset access.
|
package/docs/mcp-setup.md
CHANGED
|
@@ -1,136 +1,5 @@
|
|
|
1
1
|
# MCP / GitMCP Setup
|
|
2
2
|
|
|
3
|
-
|
|
4
|
-
|
|
5
|
-
|
|
6
|
-
|
|
7
|
-
- `mcp/servers.json`
|
|
8
|
-
- `mcp/collections.json`
|
|
9
|
-
- `mcp/fallback.json`
|
|
10
|
-
|
|
11
|
-
These are committed product config files, not hidden runtime state.
|
|
12
|
-
|
|
13
|
-
## How server mappings work
|
|
14
|
-
|
|
15
|
-
### `mcp/servers.json`
|
|
16
|
-
|
|
17
|
-
Defines the approved exact GitMCP-backed repositories the runtime may probe.
|
|
18
|
-
|
|
19
|
-
Each entry should include:
|
|
20
|
-
- `category`
|
|
21
|
-
- `repo`
|
|
22
|
-
- `endpoint`
|
|
23
|
-
|
|
24
|
-
Example:
|
|
25
|
-
|
|
26
|
-
```json
|
|
27
|
-
{
|
|
28
|
-
"category": "security",
|
|
29
|
-
"repo": "sbilly/awesome-security",
|
|
30
|
-
"endpoint": "https://gitmcp.io/sbilly/awesome-security"
|
|
31
|
-
}
|
|
32
|
-
```
|
|
33
|
-
|
|
34
|
-
### `mcp/collections.json`
|
|
35
|
-
|
|
36
|
-
Defines the committed lane-to-category policy:
|
|
37
|
-
- which categories belong to `arch`
|
|
38
|
-
- which belong to `sage`
|
|
39
|
-
- which belong to `vet`
|
|
40
|
-
|
|
41
|
-
The current runtime probes `mcp/servers.json` directly, while `mcp/collections.json` remains the committed contract for how category routing should be understood and evolved.
|
|
42
|
-
|
|
43
|
-
### `mcp/fallback.json`
|
|
44
|
-
|
|
45
|
-
Defines the documented fallback rule for:
|
|
46
|
-
- `https://gitmcp.io/docs`
|
|
47
|
-
|
|
48
|
-
Fallback is a policy declaration, not permission to skip exact endpoints when those are already known.
|
|
49
|
-
|
|
50
|
-
## Endpoint strategy
|
|
51
|
-
|
|
52
|
-
Preferred rule:
|
|
53
|
-
- use repo-specific `https://gitmcp.io/{owner}/{repo}` endpoints
|
|
54
|
-
- avoid generic discovery when an exact approved repo is already known
|
|
55
|
-
|
|
56
|
-
This keeps recommendations scoped and auditable.
|
|
57
|
-
|
|
58
|
-
## GitMCP collection examples
|
|
59
|
-
|
|
60
|
-
Useful committed examples:
|
|
61
|
-
- `sindresorhus/awesome`
|
|
62
|
-
- `dzharii/awesome-typescript`
|
|
63
|
-
- `rust-unofficial/awesome-rust`
|
|
64
|
-
- `sbilly/awesome-security`
|
|
65
|
-
- `awesome-selfhosted/awesome-selfhosted`
|
|
66
|
-
|
|
67
|
-
## How `$sage` uses live evidence
|
|
68
|
-
|
|
69
|
-
Current `$sage` behavior:
|
|
70
|
-
1. requires `architecture_status = APPROVED`
|
|
71
|
-
2. reads `mcp/servers.json`
|
|
72
|
-
3. validates endpoint shape
|
|
73
|
-
4. derives the query from the latest architecture summary
|
|
74
|
-
5. opens a live GitMCP SSE connection when live probing is enabled
|
|
75
|
-
6. negotiates the MCP message endpoint
|
|
76
|
-
7. runs `tools/list`
|
|
77
|
-
8. runs one repo-specific tool call
|
|
78
|
-
9. writes probe metadata into `.omx/evidence/sources.json`
|
|
79
|
-
|
|
80
|
-
This is enough to provide real runtime evidence for the first configured approved source. It is not a full multi-source reasoning engine yet.
|
|
81
|
-
|
|
82
|
-
## VERIFIED vs PARTIAL vs MISSING
|
|
83
|
-
|
|
84
|
-
### `VERIFIED`
|
|
85
|
-
- at least one approved endpoint is configured
|
|
86
|
-
- a real live probe succeeds
|
|
87
|
-
- usable evidence is written
|
|
88
|
-
|
|
89
|
-
### `PARTIAL`
|
|
90
|
-
- configured evidence exists
|
|
91
|
-
- but live proof is incomplete, disabled, or degraded
|
|
92
|
-
|
|
93
|
-
### `MISSING`
|
|
94
|
-
- no valid approved source exists
|
|
95
|
-
|
|
96
|
-
Do not silently mark evidence as verified when the live source did not succeed.
|
|
97
|
-
|
|
98
|
-
## What to do when MCP is unavailable
|
|
99
|
-
|
|
100
|
-
If MCP or GitMCP is unavailable:
|
|
101
|
-
- keep evidence partial or missing
|
|
102
|
-
- record the failure explicitly
|
|
103
|
-
- do not fabricate evidence
|
|
104
|
-
- fix the endpoint or network condition first
|
|
105
|
-
- rerun `$sage`
|
|
106
|
-
|
|
107
|
-
## How to keep committed MCP config honest
|
|
108
|
-
|
|
109
|
-
- commit real or intentionally documented example endpoints only
|
|
110
|
-
- do not leave fake placeholders in production config
|
|
111
|
-
- update this file if category policy changes
|
|
112
|
-
- treat endpoint drift as a release-sensitive change
|
|
113
|
-
|
|
114
|
-
## Example setup workflow
|
|
115
|
-
|
|
116
|
-
1. edit `mcp/servers.json`
|
|
117
|
-
2. confirm categories in `mcp/collections.json`
|
|
118
|
-
3. inspect fallback policy in `mcp/fallback.json`
|
|
119
|
-
4. ensure `$arch` already approved the architecture
|
|
120
|
-
5. run:
|
|
121
|
-
|
|
122
|
-
```bash
|
|
123
|
-
ma run '$sage'
|
|
124
|
-
```
|
|
125
|
-
|
|
126
|
-
6. inspect:
|
|
127
|
-
- `.omx/evidence/sources.json`
|
|
128
|
-
- `.omx/decisions.json`
|
|
129
|
-
- `.omx/release.json`
|
|
130
|
-
|
|
131
|
-
## Related surfaces
|
|
132
|
-
|
|
133
|
-
- [README.md](../README.md)
|
|
134
|
-
- [docs/skills.md](./skills.md)
|
|
135
|
-
- [docs/release-spec.md](./release-spec.md)
|
|
136
|
-
- [plugins/meta-architect/README.md](../plugins/meta-architect/README.md)
|
|
3
|
+
1. Add repo-specific GitMCP endpoints in `mcp/servers.json`.
|
|
4
|
+
2. Confirm categories in `mcp/collections.json`.
|
|
5
|
+
3. Use `https://gitmcp.io/docs` only when no approved exact endpoint exists.
|