@zhiman_innies/innies-codex 0.122.32 → 0.122.35
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/README.md
CHANGED
|
@@ -1,60 +1,350 @@
|
|
|
1
|
-
<
|
|
2
|
-
|
|
3
|
-
<
|
|
4
|
-
<
|
|
1
|
+
<div align="center">
|
|
2
|
+
|
|
3
|
+
<picture>
|
|
4
|
+
<source media="(prefers-color-scheme: dark)" srcset=".github/innies-banner-dark.svg">
|
|
5
|
+
<img alt="Innies Codex · 知满科技 AI Coding Agent" src=".github/innies-banner.svg" width="600">
|
|
6
|
+
</picture>
|
|
7
|
+
|
|
8
|
+
<p>
|
|
9
|
+
<b>知满科技 AI 编码助手</b><br>
|
|
10
|
+
<sub>基于 OpenAI Codex 深度定制 · 模型替换为知满自研 <code>qwen35_35b</code></sub>
|
|
5
11
|
</p>
|
|
6
|
-
|
|
7
|
-
|
|
8
|
-
|
|
9
|
-
|
|
12
|
+
|
|
13
|
+
<p>
|
|
14
|
+
<img alt="license" src="https://img.shields.io/badge/license-Apache--2.0-7B5CFF?style=flat-square&labelColor=0d1117">
|
|
15
|
+
<img alt="node" src="https://img.shields.io/badge/node-%E2%89%A516-01D280?style=flat-square&labelColor=0d1117">
|
|
16
|
+
<img alt="platform" src="https://img.shields.io/badge/platform-macOS_%7C_Windows_%7C_Linux-8b949e?style=flat-square&labelColor=0d1117">
|
|
17
|
+
<img alt="model" src="https://img.shields.io/badge/model-qwen35__35b-00B8D9?style=flat-square&labelColor=0d1117">
|
|
18
|
+
<img alt="context" src="https://img.shields.io/badge/context-272k-7B5CFF?style=flat-square&labelColor=0d1117">
|
|
19
|
+
</p>
|
|
20
|
+
|
|
21
|
+
<sub>
|
|
22
|
+
<a href="#速览">速览</a> ·
|
|
23
|
+
<a href="#安装">安装</a> ·
|
|
24
|
+
<a href="#配置">配置</a> ·
|
|
25
|
+
<a href="#快速上手">快速上手</a> ·
|
|
26
|
+
<a href="#与原生-codex-的差异">差异</a> ·
|
|
27
|
+
<a href="#-迭代路线">🛣️ 路线</a> ·
|
|
28
|
+
<a href="#-质量验证现状">🧪 验证</a> ·
|
|
29
|
+
<a href="#-已知限制与风险">⚠️ 风险</a> ·
|
|
30
|
+
<a href="#文档">文档</a>
|
|
31
|
+
</sub>
|
|
32
|
+
|
|
33
|
+
</div>
|
|
10
34
|
|
|
11
35
|
---
|
|
12
36
|
|
|
13
|
-
##
|
|
37
|
+
## 速览
|
|
38
|
+
|
|
39
|
+
> [!NOTE]
|
|
40
|
+
> 把官方 Codex CLI 的"大脑"换成知满自研模型 `qwen35_35b`，骨架（Agent 编排、TUI、MCP、沙箱、SDK）原样保留。
|
|
14
41
|
|
|
15
|
-
|
|
42
|
+
| 维度 | 内容 |
|
|
43
|
+
| :--- | :--- |
|
|
44
|
+
| **为什么换** | 数据不出境 · 无需 ChatGPT 账号 · 可私有化部署到客户机房 · 与原版 Codex 配置完全隔离 |
|
|
45
|
+
| **核心取舍** | `qwen35_35b` 综合推理弱于 GPT-5 · 不支持并行工具调用 · 长链路重构需更多人工拆解 |
|
|
46
|
+
| **谁该用** | 合规优先、内网部署、可接受"能力换主权"的团队 |
|
|
47
|
+
| **谁不该用** | 追求最强模型能力、复杂长链路推理、不要求数据驻留的场景——请继续用官方 Codex / Claude Code |
|
|
48
|
+
|
|
49
|
+
---
|
|
16
50
|
|
|
17
|
-
|
|
51
|
+
## 安装
|
|
18
52
|
|
|
19
|
-
```
|
|
20
|
-
|
|
21
|
-
|
|
53
|
+
```bash
|
|
54
|
+
npm install -g @zhiman_innies/innies-codex@latest
|
|
55
|
+
innies --version
|
|
22
56
|
```
|
|
23
57
|
|
|
24
|
-
|
|
25
|
-
|
|
26
|
-
|
|
58
|
+
> 要求：Node.js ≥ 16 · macOS 12+ / Windows 10/11（**Linux 不支持**，启动时编译期强制退出）
|
|
59
|
+
>
|
|
60
|
+
> 源码构建参见 [`docs/install.md`](docs/install.md)。
|
|
61
|
+
|
|
62
|
+
---
|
|
63
|
+
|
|
64
|
+
## 配置
|
|
65
|
+
|
|
66
|
+
首次运行 `innies` 会在 `~/.inniescoder/config.toml`（Windows: `%USERPROFILE%\.inniescoder\config.toml`）写入默认配置：
|
|
67
|
+
|
|
68
|
+
```toml
|
|
69
|
+
model_provider = "zhiman_35b"
|
|
70
|
+
model = "qwen35_35b"
|
|
71
|
+
|
|
72
|
+
[model_providers.zhiman_35b]
|
|
73
|
+
name = "zhiman_35b"
|
|
74
|
+
base_url = "http://101.237.37.116:7380/v1"
|
|
75
|
+
wire_api = "responses"
|
|
76
|
+
env_key = "ZHIMAN_35B_API_KEY"
|
|
27
77
|
```
|
|
28
78
|
|
|
29
|
-
|
|
79
|
+
> [!IMPORTANT]
|
|
80
|
+
> **默认 `base_url` 仅用于评估与 POC**。
|
|
81
|
+
> 正式落地通常将模型私有化部署到客户机房内网，部署完成后请**新增**一个供应商节并把 `model_provider` 指向它（默认段会被启动器周期性校正，请勿直接改）：
|
|
30
82
|
|
|
31
|
-
|
|
32
|
-
|
|
83
|
+
```toml
|
|
84
|
+
model_provider = "my_zhiman"
|
|
85
|
+
model = "qwen35_35b"
|
|
33
86
|
|
|
34
|
-
|
|
87
|
+
[model_providers.my_zhiman]
|
|
88
|
+
name = "my_zhiman"
|
|
89
|
+
base_url = "https://your-internal-host/v1" # 内网地址，由实施团队提供
|
|
90
|
+
wire_api = "responses"
|
|
91
|
+
env_key = "ZHIMAN_35B_API_KEY"
|
|
92
|
+
```
|
|
35
93
|
|
|
36
|
-
|
|
37
|
-
- Apple Silicon/arm64: `codex-aarch64-apple-darwin.tar.gz`
|
|
38
|
-
- x86_64 (older Mac hardware): `codex-x86_64-apple-darwin.tar.gz`
|
|
39
|
-
- Linux
|
|
40
|
-
- x86_64: `codex-x86_64-unknown-linux-musl.tar.gz`
|
|
41
|
-
- arm64: `codex-aarch64-unknown-linux-musl.tar.gz`
|
|
94
|
+
API Key 通过环境变量注入，**不要**写进 `config.toml`：
|
|
42
95
|
|
|
43
|
-
|
|
96
|
+
```bash
|
|
97
|
+
export ZHIMAN_35B_API_KEY="你的密钥" # 写入 ~/.zshrc 或 ~/.bashrc 后持久化
|
|
98
|
+
```
|
|
44
99
|
|
|
45
|
-
|
|
100
|
+
完整步骤：[`docs/Inniescoder用户使用手册.md`](docs/Inniescoder用户使用手册.md)
|
|
46
101
|
|
|
47
|
-
|
|
102
|
+
---
|
|
48
103
|
|
|
49
|
-
|
|
104
|
+
## 快速上手
|
|
105
|
+
|
|
106
|
+
```bash
|
|
107
|
+
cd /path/to/your/project
|
|
108
|
+
|
|
109
|
+
innies # 交互式 TUI
|
|
110
|
+
innies exec "运行所有测试并总结失败用例" # 非交互执行，适合 CI
|
|
111
|
+
innies review --base main # 代码审查
|
|
112
|
+
innies app-server # JSON-RPC + WebSocket，供 IDE/系统集成
|
|
113
|
+
```
|
|
114
|
+
|
|
115
|
+
---
|
|
116
|
+
|
|
117
|
+
## 与原生 Codex 的差异
|
|
118
|
+
|
|
119
|
+
<table>
|
|
120
|
+
<thead>
|
|
121
|
+
<tr>
|
|
122
|
+
<th width="180">维度</th>
|
|
123
|
+
<th width="320">Innies Codex</th>
|
|
124
|
+
<th width="280">官方 OpenAI Codex</th>
|
|
125
|
+
</tr>
|
|
126
|
+
</thead>
|
|
127
|
+
<tbody>
|
|
128
|
+
<tr>
|
|
129
|
+
<td><b>默认模型</b></td>
|
|
130
|
+
<td><code>qwen35_35b</code>（知满自研 35B）</td>
|
|
131
|
+
<td><code>gpt-5.4</code> / <code>gpt-5.3-codex</code></td>
|
|
132
|
+
</tr>
|
|
133
|
+
<tr>
|
|
134
|
+
<td><b>数据流向</b></td>
|
|
135
|
+
<td>知满网关（评估）/ 客户机房内网（生产）</td>
|
|
136
|
+
<td>OpenAI 海外</td>
|
|
137
|
+
</tr>
|
|
138
|
+
<tr>
|
|
139
|
+
<td><b>接入门槛</b></td>
|
|
140
|
+
<td>API Key</td>
|
|
141
|
+
<td>ChatGPT 订阅或 OpenAI API</td>
|
|
142
|
+
</tr>
|
|
143
|
+
<tr>
|
|
144
|
+
<td><b>配置目录</b></td>
|
|
145
|
+
<td><code>~/.inniescoder/</code>（独立）</td>
|
|
146
|
+
<td><code>~/.codex/</code></td>
|
|
147
|
+
</tr>
|
|
148
|
+
<tr>
|
|
149
|
+
<td><b>TUI 品牌</b></td>
|
|
150
|
+
<td>Zhiman ASCII 字 + 中文文案</td>
|
|
151
|
+
<td>OpenAI 原版</td>
|
|
152
|
+
</tr>
|
|
153
|
+
<tr>
|
|
154
|
+
<td><b>备选模型</b></td>
|
|
155
|
+
<td>DashScope <code>qwen3.6-plus</code>（百炼）</td>
|
|
156
|
+
<td>—</td>
|
|
157
|
+
</tr>
|
|
158
|
+
<tr>
|
|
159
|
+
<td><b>稳定性补丁</b></td>
|
|
160
|
+
<td>qwen streaming · 命令中断 · 首启报错等 10+ 项</td>
|
|
161
|
+
<td>—</td>
|
|
162
|
+
</tr>
|
|
163
|
+
<tr>
|
|
164
|
+
<td><b>Agent / MCP / 沙箱 / SDK</b></td>
|
|
165
|
+
<td colspan="2" align="center"><b>与上游完全一致</b></td>
|
|
166
|
+
</tr>
|
|
167
|
+
</tbody>
|
|
168
|
+
</table>
|
|
169
|
+
|
|
170
|
+
> [!TIP]
|
|
171
|
+
> **设计原则**：*只换大脑，不换骨架。* 上游 Codex 每个 release 都会跟进，定制点仅落在模型层、品牌层、稳定性补丁层。
|
|
172
|
+
|
|
173
|
+
---
|
|
174
|
+
|
|
175
|
+
## 🛣️ 迭代路线
|
|
176
|
+
|
|
177
|
+
| 阶段 | 状态 | 内容 |
|
|
178
|
+
| :--- | :---: | :--- |
|
|
179
|
+
| **v1 · CLI Agent** | ✅ 已发布 | 稳定的 TUI / `exec` / `review` / `app-server` 入口，知满 `qwen35_35b` 适配，14 项 Superpowers 技能 |
|
|
180
|
+
| **v2 · 私有化部署套件** | 🟢 进行中 | 推理服务镜像、Helm Chart、模型权重交付、客户机房一键部署 |
|
|
181
|
+
| **v3 · InniesCoding 端到端工作流** | 🟡 内测 | 由自然语言需求自动驱动的五阶段管线，详见下文 |
|
|
182
|
+
| **v4 · IDE 插件 / Web 控制台** | ⚪ 规划中 | 基于现有 App Server 能力延伸 |
|
|
183
|
+
|
|
184
|
+
### InniesCoding 端到端工作流（v3 预览）
|
|
185
|
+
|
|
186
|
+
`InniesCoding` 是建在 Innies Codex 之上的**研发自动化管线**，由 RAG（检索增强）与 KG（知识图谱）MCP server 驱动，将一个自然语言需求拆解为五个可追溯阶段：
|
|
187
|
+
|
|
188
|
+
```
|
|
189
|
+
需求分析 → 代码定位与影响评估 → Spec 生成 → 代码变更 → 测试用例补全
|
|
190
|
+
```
|
|
191
|
+
|
|
192
|
+
| 阶段 | 输入 | 输出 |
|
|
193
|
+
| :--- | :--- | :--- |
|
|
194
|
+
| `requirement_analysis` | 原始需求 | PRD 草稿（含 RAG 证据 + KG 影响范围） |
|
|
195
|
+
| `code_search_and_assessment` | PRD + 项目代码 | 受影响文件、调用链、风险点 |
|
|
196
|
+
| `spec_generation` | 影响范围 | 技术 Spec |
|
|
197
|
+
| `code_change` | Spec | 代码补丁 |
|
|
198
|
+
| `test_case_lookup_and_generation` | 变更 | 单元 / 集成测试用例 |
|
|
199
|
+
|
|
200
|
+
> [!NOTE]
|
|
201
|
+
> 当前以 `INNIES_CODING_STAGE` 环境变量内部触发，公开版本将以 `innies coding <stage>` 子命令形式暴露。实现入口见 [`codex-cli/bin/innies-coding-runtime.js`](codex-cli/bin/innies-coding-runtime.js)。
|
|
202
|
+
|
|
203
|
+
---
|
|
204
|
+
|
|
205
|
+
## 🧪 质量验证现状
|
|
206
|
+
|
|
207
|
+
> [!NOTE]
|
|
208
|
+
> 基于 [`docs/innies-qwen35-*`](docs/) 系列报告的真实压测产出。最近一次验收：**2026-05-24**。
|
|
209
|
+
|
|
210
|
+
### 已完成
|
|
211
|
+
|
|
212
|
+
| 验证项 | 结果 | 证据 |
|
|
213
|
+
| :--- | :---: | :--- |
|
|
214
|
+
| 全链路功能矩阵（CLI / API / Skills / Output Schema / Spawn Agent） | ✅ | [full-chain 2026-05-17](docs/innies-qwen35-full-chain-test-report-2026-05-17.md) |
|
|
215
|
+
| 单用户基线 | ✅ | [single-user 2026-05-17](docs/innies-qwen35-single-user-test-report-2026-05-17.md) |
|
|
216
|
+
| 30 并发 API 突发 | ✅ 30/30 | full-chain 2026-05-17 |
|
|
217
|
+
| **复杂任务通过率（A 计划验收）** | **✅ 118/120 = 98.3%** | [complex-task-fix-report 2026-05-24](docs/innies-qwen35-complex-task-fix-report-2026-05-24.md) |
|
|
218
|
+
| · `multi_agent_review` | ✅ 40/40 = 100%（60% → 100%） | 同上 |
|
|
219
|
+
| · `system_design` | ✅ 38/40 = 95%（75% → 95%） | 同上 |
|
|
220
|
+
| · `superpowers_chain` | ✅ 40/40 = 100% | 同上 |
|
|
221
|
+
| 长稳 harness + 监控脚本就绪 | ✅ | [`scripts/long_soak.py`](scripts/long_soak.py) · [`scripts/long_soak_monitor.py`](scripts/long_soak_monitor.py) |
|
|
222
|
+
| 长稳预演（1h 子集） | ✅ | corpus `long-soak-preflight-20260524-*` |
|
|
223
|
+
| **8 小时长稳正式跑（B 计划验收）** | **✅ 302/303 = 99.67%** | [long-soak-report 2026-05-26](docs/innies-qwen35-long-soak-report-2026-05-26.md) |
|
|
224
|
+
| · `superpowers_chain` | ✅ 27/27 = 100% | 同上 |
|
|
225
|
+
| · `multi_agent_review` | ✅ 27/27 = 100% | 同上 |
|
|
226
|
+
| · `system_design` | ✅ 26/27 = 96.3%（1 次超时，无熔断） | 同上 |
|
|
227
|
+
| · KV cache 命中率 | ✅ 83.6%（37.4M input / 31.3M cached） | 同上 |
|
|
228
|
+
| 失败语料库（供后续 SFT/DPO） | ✅ 4 个样本 | [`docs/qwen35-finetune-corpus/`](docs/qwen35-finetune-corpus/) |
|
|
229
|
+
|
|
230
|
+
### 进行中 / 未完成
|
|
231
|
+
|
|
232
|
+
| 验证项 | 状态 | 说明 |
|
|
233
|
+
| :--- | :---: | :--- |
|
|
234
|
+
| 100+ 并发 API 突发 | ❌ 已知失败 | 43/102，45s 超时；待服务端容量整改后复跑 |
|
|
235
|
+
| 长稳复跑（次日不同时段） | ⚪ 未启动 | B 计划首跑已验收，复跑视客户落地节奏安排 |
|
|
236
|
+
| 24h / 7×24 持续运行 | ⚪ 未启动 | 8h 已达标，24h 验证服务长尾稳定性 |
|
|
237
|
+
| Windows 平台行为 | ⚪ 未验证 | 当前全部测试仅在 macOS 跑过；Linux 已编译期强制禁用 |
|
|
238
|
+
| 客户机房私有化部署后的长稳 | ⚪ 未验证 | 实施时与客户联合验证 |
|
|
239
|
+
| 多机器分布式 agent / 故障注入 / 多租户 | ⚪ 未启动 | B 计划延伸事项 |
|
|
240
|
+
|
|
241
|
+
### 已落地的工程工件（A 计划，2026-05-24）
|
|
242
|
+
|
|
243
|
+
- 失败归因脚本 [`scripts/classify_complex_failures.py`](scripts/classify_complex_failures.py)
|
|
244
|
+
- output_schema 压测 [`scripts/schema_pressure.py`](scripts/schema_pressure.py)
|
|
245
|
+
- 长稳 harness [`scripts/long_soak.py`](scripts/long_soak.py)（419 行）
|
|
246
|
+
- 长稳监控 [`scripts/long_soak_monitor.py`](scripts/long_soak_monitor.py)（含 RSS / FD / 目录大小 / 崩溃信号）
|
|
247
|
+
- 三路径修复（B1 Prompt 末尾固化 marker · B2 强制 output_schema · B3 Harness 层 re-prompt 兜底）合并进主压测 runner
|
|
248
|
+
|
|
249
|
+
---
|
|
250
|
+
|
|
251
|
+
## ⚠️ 已知限制与风险
|
|
252
|
+
|
|
253
|
+
> [!WARNING]
|
|
254
|
+
> 以下内容**基于代码与配置事实**列出，落地前请完整阅读。
|
|
255
|
+
|
|
256
|
+
### 1. 模型能力差距是真实存在的
|
|
257
|
+
|
|
258
|
+
`qwen35_35b` 在以下维度**可被代码直接证实**地弱于 GPT-5 系列：
|
|
259
|
+
|
|
260
|
+
<table>
|
|
261
|
+
<thead>
|
|
262
|
+
<tr>
|
|
263
|
+
<th width="220">能力</th>
|
|
264
|
+
<th width="280"><code>qwen35_35b</code></th>
|
|
265
|
+
<th width="280"><code>gpt-5.4</code> / Claude 4.7</th>
|
|
266
|
+
</tr>
|
|
267
|
+
</thead>
|
|
268
|
+
<tbody>
|
|
269
|
+
<tr>
|
|
270
|
+
<td>并行工具调用</td>
|
|
271
|
+
<td>❌ <code>supports_parallel_tool_calls: false</code></td>
|
|
272
|
+
<td>✅</td>
|
|
273
|
+
</tr>
|
|
274
|
+
<tr>
|
|
275
|
+
<td>推理强度档位</td>
|
|
276
|
+
<td>❌ <code>supported_reasoning_levels: []</code>（无 high/medium/low）</td>
|
|
277
|
+
<td>✅</td>
|
|
278
|
+
</tr>
|
|
279
|
+
<tr>
|
|
280
|
+
<td>输出截断阈值</td>
|
|
281
|
+
<td>10 000 tokens（较激进）</td>
|
|
282
|
+
<td>同</td>
|
|
283
|
+
</tr>
|
|
284
|
+
<tr>
|
|
285
|
+
<td>综合代码任务</td>
|
|
286
|
+
<td colspan="2">主观体验：复杂多文件重构、长链路调试明显落后</td>
|
|
287
|
+
</tr>
|
|
288
|
+
</tbody>
|
|
289
|
+
</table>
|
|
290
|
+
|
|
291
|
+
**实际影响**：
|
|
292
|
+
|
|
293
|
+
- 一次 turn 里需要"读 5 个文件 + grep 3 次"时，`qwen35_35b` 会**串行**逐个发起，整体延迟约为 GPT-5 的 2-3 倍
|
|
294
|
+
- 跨 10+ 文件的架构级重构需要更多人工拆解和多轮迭代，**不要期望一次到位**
|
|
295
|
+
- 不能通过 `model_reasoning_effort` 让模型"想久一点"——这个开关对 `qwen35_35b` 无效
|
|
296
|
+
|
|
297
|
+
> [!TIP]
|
|
298
|
+
> **已工程化缓解**：复杂任务里"指令跟随不稳定"（如 final marker 缺失）这类故障，已通过 A 计划落地的三路径修复（Prompt 末尾固化 marker + 强制 output_schema + Harness re-prompt 兜底）将 `multi_agent_review` 从 60% 提升到 100%、`system_design` 从 75% 提升到 95%。详见 [质量验证现状](#-质量验证现状)。模型本身的并行/推理档位差距仍然存在。
|
|
299
|
+
|
|
300
|
+
### 2. 上游跟随有滞后与冲突风险
|
|
301
|
+
|
|
302
|
+
- 仓库跟随官方 Codex 每个 release（最近一次 `rust-v0.122.32`）
|
|
303
|
+
- 上游做 breaking change 时，定制点（品牌、模型目录、隔离逻辑）需要手动 cherry-pick，**预期延迟 1~2 周**
|
|
304
|
+
- 上游 OpenAI 一旦移除 Responses API 或调整工具调用 schema，`qwen35_35b` 适配需追加工作量
|
|
305
|
+
|
|
306
|
+
### 3. 治理边界仅覆盖到本机
|
|
307
|
+
|
|
308
|
+
- macOS 沙箱基于 Seatbelt（`sandbox-exec`），Windows 沙箱能力受限（仅文件系统 ACL）；**Linux 不在支持范围内**，启动时即报错退出
|
|
309
|
+
- 没有组织级 IAM、角色、审计后台——多人协作的合规治理需对接外部系统
|
|
310
|
+
- API Key 是单 token、无作用域、无过期——一旦泄露需手动轮换全员
|
|
311
|
+
|
|
312
|
+
### 4. 许可与权属
|
|
313
|
+
|
|
314
|
+
- 仓库代码遵循 **Apache-2.0**（见 [`LICENSE`](LICENSE)）
|
|
315
|
+
- `qwen35_35b` 模型权重**不在本仓库**，使用知满推理服务即受其商务条款约束；客户机房私有化部署的权重交付由知满实施团队处理
|
|
316
|
+
- TUI 品牌字、`Zhiman` 商标属知满科技
|
|
317
|
+
|
|
318
|
+
---
|
|
319
|
+
|
|
320
|
+
## 文档
|
|
321
|
+
|
|
322
|
+
| 文档 | 内容 |
|
|
323
|
+
| :--- | :--- |
|
|
324
|
+
| [`docs/Inniescoder用户使用手册.md`](docs/Inniescoder用户使用手册.md) | 面向最终用户的安装 / 配置 / 升级 / 卸载 / FAQ |
|
|
325
|
+
| [`docs/Inniescoder能力支持说明.md`](docs/Inniescoder能力支持说明.md) | 代码辅助能力逐项对照表 |
|
|
326
|
+
| [`docs/install.md`](docs/install.md) | 从源码构建 |
|
|
327
|
+
| [`docs/config.md`](docs/config.md) | `config.toml` 字段参考 |
|
|
328
|
+
| [`docs/sandbox.md`](docs/sandbox.md) | 沙箱与审批策略 |
|
|
329
|
+
| [`docs/skills.md`](docs/skills.md) | Skill 注入机制 |
|
|
330
|
+
| [`docs/innies-qwen35-complex-task-fix-plan.md`](docs/innies-qwen35-complex-task-fix-plan.md) | **A 计划**：复杂任务成功率整改方案 |
|
|
331
|
+
| [`docs/innies-qwen35-complex-task-fix-report-2026-05-24.md`](docs/innies-qwen35-complex-task-fix-report-2026-05-24.md) | **A 计划验收报告**（98.3% 通过率） |
|
|
332
|
+
| [`docs/innies-qwen35-long-soak-plan.md`](docs/innies-qwen35-long-soak-plan.md) | **B 计划**：8 小时长稳运行方案 |
|
|
333
|
+
| [`docs/innies-qwen35-long-soak-report-2026-05-26.md`](docs/innies-qwen35-long-soak-report-2026-05-26.md) | **B 计划验收报告**（302/303 = 99.67% 通过率） |
|
|
334
|
+
| [`docs/innies-qwen35-high-concurrency-plan.md`](docs/innies-qwen35-high-concurrency-plan.md) | 高并发整改计划（多 key 池 / 多端点 / 熔断） |
|
|
335
|
+
| [`docs/innies-qwen35-stability-test-report-2026-05-19.md`](docs/innies-qwen35-stability-test-report-2026-05-19.md) | 修复前稳定性基线报告 |
|
|
336
|
+
| [`docs/innies-qwen35-*.md`](docs/) | 其他 qwen35 全链路 / 单用户 / 早期报告 |
|
|
337
|
+
| [`AGENTS.md`](AGENTS.md) | 贡献者必读（Rust 规范、测试约定） |
|
|
338
|
+
| [`SECURITY.md`](SECURITY.md) | 安全披露流程 |
|
|
339
|
+
|
|
340
|
+
---
|
|
50
341
|
|
|
51
|
-
|
|
342
|
+
<div align="center">
|
|
52
343
|
|
|
53
|
-
|
|
344
|
+
<sub>· ✦ ·</sub>
|
|
54
345
|
|
|
55
|
-
|
|
56
|
-
|
|
57
|
-
|
|
58
|
-
- [**Open source fund**](./docs/open-source-fund.md)
|
|
346
|
+
<sub>问题反馈 · 定制需求 → <a href="https://zhiman.tech/"><b>知满科技</b></a></sub>
|
|
347
|
+
<br>
|
|
348
|
+
<sub><i>Zhiman · intent in motion</i></sub>
|
|
59
349
|
|
|
60
|
-
|
|
350
|
+
</div>
|
|
@@ -17,6 +17,19 @@ Do NOT invoke any implementation skill, write any code, scaffold any project, or
|
|
|
17
17
|
|
|
18
18
|
Every project goes through this process. A todo list, a single-function utility, a config change — all of them. "Simple" projects are where unexamined assumptions cause the most wasted work. The design can be short (a few sentences for truly simple projects), but you MUST present it and get approval.
|
|
19
19
|
|
|
20
|
+
## Anti-Pattern: "This Is Conceptual, Not An Implementation Task"
|
|
21
|
+
|
|
22
|
+
STOP. If you find yourself thinking any of the following — you are rationalizing your way out of the skill, and you MUST NOT act on the thought:
|
|
23
|
+
|
|
24
|
+
- "This is a conceptual question, not a code task — I should just answer directly."
|
|
25
|
+
- "The user is asking how X works / how to design X — they want an explanation, not collaboration."
|
|
26
|
+
- "Exploration is unnecessary for a design discussion — I'll just lay out the standard answer."
|
|
27
|
+
- "I can describe the three-layer / N-tier / standard architecture from memory; brainstorming is overkill."
|
|
28
|
+
|
|
29
|
+
Brainstorming exists **precisely** for design conversations. "How should I design X" is the canonical case, not the exception. The user said "我要设计 / I want to design / let's build" — that is an explicit invitation to collaborate, not a request for a textbook answer.
|
|
30
|
+
|
|
31
|
+
The correct first move is **always** the checklist below, starting with project context exploration and clarifying questions. Jumping straight to "## 1. Short-term memory ## 2. Long-term memory ## 3. Working memory" with no questions asked is the failure mode this skill is designed to prevent.
|
|
32
|
+
|
|
20
33
|
## Checklist
|
|
21
34
|
|
|
22
35
|
You MUST create a task for each of these items and complete them in order:
|
|
@@ -93,6 +93,9 @@ These thoughts mean STOP—you're rationalizing:
|
|
|
93
93
|
| "I'll just do this one thing first" | Check BEFORE doing anything. |
|
|
94
94
|
| "This feels productive" | Undisciplined action wastes time. Skills prevent this. |
|
|
95
95
|
| "I know what that means" | Knowing the concept ≠ using the skill. Invoke it. |
|
|
96
|
+
| **"This is conceptual, not implementation"** | **Brainstorming IS for conceptual design talk. "How should I design X" is the canonical case, not the exception.** |
|
|
97
|
+
| **"The user wants an explanation, not collaboration"** | **If they wrote "我要设计 / I want to design / let's build" they invited collaboration. Treat it as such.** |
|
|
98
|
+
| **"I can give the standard textbook answer from memory"** | **Standard answers without context-aware clarification are exactly what brainstorming is designed to replace.** |
|
|
96
99
|
|
|
97
100
|
## Skill Priority
|
|
98
101
|
|
package/bin/innies.js
CHANGED
|
File without changes
|
package/package.json
CHANGED
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
{
|
|
2
2
|
"name": "@zhiman_innies/innies-codex",
|
|
3
|
-
"version": "0.122.
|
|
3
|
+
"version": "0.122.35",
|
|
4
4
|
"license": "Apache-2.0",
|
|
5
5
|
"bin": {
|
|
6
6
|
"innies": "bin/innies.js"
|
|
@@ -23,9 +23,9 @@
|
|
|
23
23
|
"postinstall": "node bin/innies-init.js"
|
|
24
24
|
},
|
|
25
25
|
"optionalDependencies": {
|
|
26
|
-
"@zhiman_innies/innies-codex-darwin-x64": "0.122.
|
|
27
|
-
"@zhiman_innies/innies-codex-darwin-arm64": "0.122.
|
|
28
|
-
"@zhiman_innies/innies-codex-win32-x64": "0.122.
|
|
29
|
-
"@zhiman_innies/innies-codex-win32-arm64": "0.122.
|
|
26
|
+
"@zhiman_innies/innies-codex-darwin-x64": "0.122.35-darwin-x64",
|
|
27
|
+
"@zhiman_innies/innies-codex-darwin-arm64": "0.122.35-darwin-arm64",
|
|
28
|
+
"@zhiman_innies/innies-codex-win32-x64": "0.122.35-win32-x64",
|
|
29
|
+
"@zhiman_innies/innies-codex-win32-arm64": "0.122.35-win32-arm64"
|
|
30
30
|
}
|
|
31
31
|
}
|