@yeongjaeyou/claude-code-config 0.4.0 → 0.5.0
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/commands/ask-codex.md +131 -345
- package/.claude/commands/ask-deepwiki.md +15 -15
- package/.claude/commands/ask-gemini.md +134 -352
- package/.claude/commands/code-review.md +41 -40
- package/.claude/commands/commit-and-push.md +35 -36
- package/.claude/commands/council.md +318 -0
- package/.claude/commands/edit-notebook.md +34 -33
- package/.claude/commands/gh/create-issue-label.md +19 -17
- package/.claude/commands/gh/decompose-issue.md +66 -65
- package/.claude/commands/gh/init-project.md +46 -52
- package/.claude/commands/gh/post-merge.md +74 -79
- package/.claude/commands/gh/resolve-issue.md +38 -46
- package/.claude/commands/plan.md +15 -14
- package/.claude/commands/tm/convert-prd.md +53 -53
- package/.claude/commands/tm/post-merge.md +92 -112
- package/.claude/commands/tm/resolve-issue.md +148 -154
- package/.claude/commands/tm/review-prd-with-codex.md +272 -279
- package/.claude/commands/tm/sync-to-github.md +189 -212
- package/.claude/guidelines/cv-guidelines.md +30 -0
- package/.claude/guidelines/id-reference.md +34 -0
- package/.claude/guidelines/work-guidelines.md +17 -0
- package/.claude/skills/notion-md-uploader/SKILL.md +252 -0
- package/.claude/skills/notion-md-uploader/references/notion_block_types.md +323 -0
- package/.claude/skills/notion-md-uploader/references/setup_guide.md +156 -0
- package/.claude/skills/notion-md-uploader/scripts/__pycache__/markdown_parser.cpython-311.pyc +0 -0
- package/.claude/skills/notion-md-uploader/scripts/__pycache__/notion_client.cpython-311.pyc +0 -0
- package/.claude/skills/notion-md-uploader/scripts/__pycache__/notion_converter.cpython-311.pyc +0 -0
- package/.claude/skills/notion-md-uploader/scripts/markdown_parser.py +607 -0
- package/.claude/skills/notion-md-uploader/scripts/notion_client.py +337 -0
- package/.claude/skills/notion-md-uploader/scripts/notion_converter.py +477 -0
- package/.claude/skills/notion-md-uploader/scripts/upload_md.py +298 -0
- package/.claude/skills/skill-creator/LICENSE.txt +202 -0
- package/.claude/skills/skill-creator/SKILL.md +209 -0
- package/.claude/skills/skill-creator/scripts/init_skill.py +303 -0
- package/.claude/skills/skill-creator/scripts/package_skill.py +110 -0
- package/.claude/skills/skill-creator/scripts/quick_validate.py +65 -0
- package/package.json +1 -1
|
@@ -1,101 +1,101 @@
|
|
|
1
|
-
# PRD Codex
|
|
1
|
+
# PRD Review with Codex (review-prd-with-codex)
|
|
2
2
|
|
|
3
|
-
|
|
4
|
-
|
|
3
|
+
Review the generated PRD using Codex MCP, with Claude performing cross-check validation to deliver a consensus conclusion.
|
|
4
|
+
Ping-pong up to 3 times until consensus is reached.
|
|
5
5
|
|
|
6
|
-
|
|
7
|
-
- Codex
|
|
8
|
-
-
|
|
9
|
-
-
|
|
10
|
-
-
|
|
6
|
+
**Core Principles:**
|
|
7
|
+
- Codex has limited context/tools, so Claude must verify
|
|
8
|
+
- On disagreement, re-query Codex (with context re-transmission)
|
|
9
|
+
- No emoji usage
|
|
10
|
+
- Use generic expressions (avoid project-specific terminology)
|
|
11
11
|
|
|
12
12
|
---
|
|
13
13
|
|
|
14
|
-
##
|
|
14
|
+
## Arguments
|
|
15
15
|
|
|
16
|
-
`$ARGUMENTS
|
|
17
|
-
-
|
|
18
|
-
-
|
|
16
|
+
`$ARGUMENTS` receives the PRD file path.
|
|
17
|
+
- Example: `/tm:review-prd-with-codex .taskmaster/docs/prd.md`
|
|
18
|
+
- If missing, request path input via AskUserQuestion
|
|
19
19
|
|
|
20
20
|
---
|
|
21
21
|
|
|
22
|
-
## Codex
|
|
23
|
-
|
|
24
|
-
### 1.
|
|
25
|
-
-
|
|
26
|
-
- sandbox
|
|
27
|
-
-
|
|
28
|
-
-
|
|
29
|
-
-
|
|
30
|
-
|
|
31
|
-
### 2.
|
|
32
|
-
- PRD
|
|
33
|
-
- **CLAUDE.md
|
|
34
|
-
-
|
|
35
|
-
-
|
|
36
|
-
- **PRD
|
|
37
|
-
-
|
|
38
|
-
-
|
|
39
|
-
|
|
40
|
-
### 3.
|
|
41
|
-
- Claude
|
|
42
|
-
-
|
|
43
|
-
|
|
44
|
-
### 4.
|
|
45
|
-
- **CLAUDE.md
|
|
46
|
-
-
|
|
47
|
-
-
|
|
48
|
-
-
|
|
49
|
-
|
|
50
|
-
### 5.
|
|
51
|
-
-
|
|
52
|
-
- "
|
|
53
|
-
-
|
|
22
|
+
## Codex Context Configuration Principles
|
|
23
|
+
|
|
24
|
+
### 1. Codex Usage Scope for This Workflow
|
|
25
|
+
- This command uses **Codex MCP tool** (`mcp__codex__codex`)
|
|
26
|
+
- Set sandbox to `read-only`
|
|
27
|
+
- **No network/web search** (for reproducibility and safety)
|
|
28
|
+
- Include PRD/context directly in prompt for reproducibility (don't rely on file exploration)
|
|
29
|
+
- Git history not accessible (Claude cross-check compensates)
|
|
30
|
+
|
|
31
|
+
### 2. Required Information (Rich Context)
|
|
32
|
+
- Full PRD text (with line numbers - `nl -ba` format)
|
|
33
|
+
- **Full CLAUDE.md** (include entire content without limits)
|
|
34
|
+
- **Detailed project tech stack** (package.json, requirements.txt contents)
|
|
35
|
+
- **Directory structure** (main folder layout)
|
|
36
|
+
- **PRD-related code summary** (optional: symbol overview of files mentioned in PRD)
|
|
37
|
+
- Explicit review criteria
|
|
38
|
+
- Enforced output format
|
|
39
|
+
|
|
40
|
+
### 3. Information to Exclude
|
|
41
|
+
- Claude-specific tools (AskUserQuestion, TodoWrite, etc.)
|
|
42
|
+
- Internal workflow details
|
|
43
|
+
|
|
44
|
+
### 4. Prompt Construction Principles
|
|
45
|
+
- **Include full CLAUDE.md** (remove 1000 character limit)
|
|
46
|
+
- Review entire PRD at once (don't split)
|
|
47
|
+
- Provide project context with tech stack and directory structure
|
|
48
|
+
- Consider prompt reduction only on timeout
|
|
49
|
+
|
|
50
|
+
### 5. Enforced Output Format
|
|
51
|
+
- Specify structured format (tables, lists)
|
|
52
|
+
- State "Must use this format"
|
|
53
|
+
- Require line number references
|
|
54
54
|
|
|
55
55
|
---
|
|
56
56
|
|
|
57
|
-
##
|
|
57
|
+
## Workflow Steps
|
|
58
58
|
|
|
59
|
-
### Step 1:
|
|
59
|
+
### Step 1: Gather Pre-requisite Information (Rich Context)
|
|
60
60
|
|
|
61
|
-
#### 1.1 PRD
|
|
61
|
+
#### 1.1 Read PRD File (with line numbers)
|
|
62
62
|
```bash
|
|
63
63
|
nl -ba $ARGUMENTS
|
|
64
64
|
```
|
|
65
|
-
- `nl -ba`:
|
|
66
|
-
-
|
|
65
|
+
- `nl -ba`: Numbers all lines including blank lines (more consistent than cat -n)
|
|
66
|
+
- Output error message and exit if file doesn't exist
|
|
67
67
|
|
|
68
|
-
#### 1.2 PRD
|
|
68
|
+
#### 1.2 Verify PRD File Exists
|
|
69
69
|
```bash
|
|
70
|
-
test -f $ARGUMENTS && echo "
|
|
70
|
+
test -f $ARGUMENTS && echo "File exists" || echo "File not found"
|
|
71
71
|
```
|
|
72
72
|
|
|
73
|
-
#### 1.3 CLAUDE.md
|
|
74
|
-
-
|
|
75
|
-
-
|
|
76
|
-
-
|
|
73
|
+
#### 1.3 Read Full CLAUDE.md
|
|
74
|
+
- Check CLAUDE.md in project root
|
|
75
|
+
- **Include entire content in prompt** (no 1000 character limit)
|
|
76
|
+
- Convey all project conventions, tech stack, and caveats
|
|
77
77
|
|
|
78
|
-
#### 1.4
|
|
79
|
-
Claude
|
|
80
|
-
- `package.json` (
|
|
81
|
-
- `requirements.txt` (
|
|
82
|
-
- `go.mod`, `Cargo.toml
|
|
83
|
-
-
|
|
78
|
+
#### 1.4 Collect Detailed Project Tech Stack
|
|
79
|
+
Claude reads and summarizes these files:
|
|
80
|
+
- `package.json` (frontend dependencies)
|
|
81
|
+
- `requirements.txt` (backend dependencies)
|
|
82
|
+
- `go.mod`, `Cargo.toml`, etc. (if applicable)
|
|
83
|
+
- Specify major framework/library versions
|
|
84
84
|
|
|
85
|
-
#### 1.5
|
|
85
|
+
#### 1.5 Collect Directory Structure
|
|
86
86
|
```bash
|
|
87
|
-
#
|
|
87
|
+
# Main directory structure (depth 2-3)
|
|
88
88
|
tree -L 3 --dirsfirst -I 'node_modules|__pycache__|.git|dist|build|.next'
|
|
89
89
|
```
|
|
90
|
-
|
|
90
|
+
Or use `ls -la` combinations to understand structure
|
|
91
91
|
|
|
92
|
-
#### 1.6 PRD
|
|
93
|
-
-
|
|
94
|
-
-
|
|
95
|
-
-
|
|
92
|
+
#### 1.6 Summarize PRD-Related Code Files (Optional)
|
|
93
|
+
- If major files/modules are mentioned in PRD, collect symbol overview
|
|
94
|
+
- Example: For "Authentication System Improvement" PRD, check `routes/auth.py`, `middleware.ts`, etc.
|
|
95
|
+
- Helps understand relationship between existing implementation and PRD design
|
|
96
96
|
|
|
97
|
-
#### 1.7 TaskMaster PRD
|
|
98
|
-
`.taskmaster/templates/example_prd.txt`
|
|
97
|
+
#### 1.7 Reference TaskMaster PRD Template
|
|
98
|
+
`.taskmaster/templates/example_prd.txt` structure:
|
|
99
99
|
```
|
|
100
100
|
<context>
|
|
101
101
|
# Overview
|
|
@@ -111,29 +111,29 @@ tree -L 3 --dirsfirst -I 'node_modules|__pycache__|.git|dist|build|.next'
|
|
|
111
111
|
</PRD>
|
|
112
112
|
```
|
|
113
113
|
|
|
114
|
-
### Step 2: Codex
|
|
114
|
+
### Step 2: Construct Codex Prompt (Rich Context)
|
|
115
115
|
|
|
116
|
-
|
|
116
|
+
Use this template to write the prompt:
|
|
117
117
|
|
|
118
118
|
```
|
|
119
|
-
##
|
|
120
|
-
|
|
119
|
+
## Role
|
|
120
|
+
You are a PRD (Product Requirements Document) review expert.
|
|
121
121
|
|
|
122
|
-
##
|
|
122
|
+
## Project Context
|
|
123
123
|
|
|
124
|
-
###
|
|
125
|
-
[package.json / requirements.txt
|
|
126
|
-
|
|
124
|
+
### Tech Stack
|
|
125
|
+
[Summary of package.json / requirements.txt contents]
|
|
126
|
+
Example:
|
|
127
127
|
- Frontend: Next.js 15, React 19, Tailwind CSS v4, shadcn/ui
|
|
128
128
|
- Backend: FastAPI, Google Gemini API, Decord
|
|
129
|
-
-
|
|
129
|
+
- Deployment: Docker + Cloudflare Tunnel
|
|
130
130
|
|
|
131
|
-
###
|
|
132
|
-
[tree
|
|
131
|
+
### Project Structure
|
|
132
|
+
[tree or ls output]
|
|
133
133
|
```
|
|
134
134
|
frontend/
|
|
135
135
|
├── src/app/ # App Router
|
|
136
|
-
├── src/components/ # UI
|
|
136
|
+
├── src/components/ # UI components
|
|
137
137
|
└── src/hooks/ # Custom hooks
|
|
138
138
|
|
|
139
139
|
services/
|
|
@@ -142,308 +142,301 @@ services/
|
|
|
142
142
|
└── child_safety_processor.py
|
|
143
143
|
```
|
|
144
144
|
|
|
145
|
-
###
|
|
146
|
-
[CLAUDE.md
|
|
145
|
+
### Project Guidelines (CLAUDE.md)
|
|
146
|
+
[Full CLAUDE.md - include entire content without limits]
|
|
147
147
|
|
|
148
|
-
###
|
|
149
|
-
[
|
|
148
|
+
### Existing Code Related to PRD (Optional)
|
|
149
|
+
[Symbol overview of major files mentioned in PRD]
|
|
150
150
|
|
|
151
|
-
##
|
|
152
|
-
|
|
151
|
+
## Review Target
|
|
152
|
+
File: [PRD file path]
|
|
153
153
|
|
|
154
|
-
### PRD
|
|
155
|
-
[PRD
|
|
154
|
+
### PRD Content (with line numbers)
|
|
155
|
+
[Full PRD - in nl -ba format]
|
|
156
156
|
|
|
157
|
-
## TaskMaster PRD
|
|
157
|
+
## TaskMaster PRD Format Criteria
|
|
158
158
|
<context>: Overview, Core Features, User Experience
|
|
159
|
-
<PRD>: Technical Architecture, Development Roadmap (Phase
|
|
159
|
+
<PRD>: Technical Architecture, Development Roadmap (by Phase),
|
|
160
160
|
Logical Dependency Chain, Risks and Mitigations, Appendix
|
|
161
161
|
|
|
162
|
-
##
|
|
163
|
-
1.
|
|
164
|
-
2.
|
|
165
|
-
3.
|
|
166
|
-
4.
|
|
167
|
-
Success Metrics, Risk/Dependencies
|
|
168
|
-
5.
|
|
162
|
+
## Review Criteria
|
|
163
|
+
1. Structure: Compliance with TaskMaster PRD format
|
|
164
|
+
2. Clarity: Ambiguous expressions, undefined terms, unmeasurable goals
|
|
165
|
+
3. Feasibility: Implementation-level specificity, technical realism
|
|
166
|
+
4. Completeness: Missing sections (User Stories, Acceptance Criteria,
|
|
167
|
+
Success Metrics, Risk/Dependencies, etc.)
|
|
168
|
+
5. Consistency: Internal contradictions, duplicate definitions, version mismatches
|
|
169
169
|
|
|
170
|
-
##
|
|
170
|
+
## Output Format (Must use this format)
|
|
171
171
|
|
|
172
|
-
###
|
|
173
|
-
- [
|
|
172
|
+
### Strengths
|
|
173
|
+
- [Item]: [Description] (line number)
|
|
174
174
|
|
|
175
|
-
###
|
|
176
|
-
|
|
|
177
|
-
|
|
175
|
+
### Issues
|
|
176
|
+
| Item | Location (line) | Problem | Recommended Fix |
|
|
177
|
+
|------|-----------------|---------|-----------------|
|
|
178
178
|
|
|
179
|
-
###
|
|
180
|
-
- [
|
|
181
|
-
- [
|
|
179
|
+
### Open Questions
|
|
180
|
+
- [Question 1]
|
|
181
|
+
- [Question 2]
|
|
182
182
|
|
|
183
|
-
###
|
|
184
|
-
[1-2
|
|
183
|
+
### Overall Assessment
|
|
184
|
+
[1-2 sentence summary]
|
|
185
185
|
```
|
|
186
186
|
|
|
187
|
-
### Step 3: Codex MCP
|
|
187
|
+
### Step 3: Execute Codex MCP (First Run)
|
|
188
188
|
|
|
189
|
-
**mcp__codex__codex
|
|
189
|
+
**Call mcp__codex__codex tool:**
|
|
190
190
|
|
|
191
|
-
|
|
192
|
-
- `prompt`:
|
|
191
|
+
Parameters:
|
|
192
|
+
- `prompt`: Full prompt constructed in Step 2
|
|
193
193
|
- `sandbox`: "read-only"
|
|
194
|
-
- `cwd`:
|
|
194
|
+
- `cwd`: Project root path (optional)
|
|
195
195
|
|
|
196
|
-
|
|
196
|
+
**Call example:**
|
|
197
197
|
```
|
|
198
|
-
mcp__codex__codex
|
|
199
|
-
- prompt: [
|
|
198
|
+
mcp__codex__codex tool call:
|
|
199
|
+
- prompt: [constructed prompt]
|
|
200
200
|
- sandbox: "read-only"
|
|
201
201
|
```
|
|
202
202
|
|
|
203
|
-
|
|
204
|
-
-
|
|
205
|
-
-
|
|
203
|
+
**Notes:**
|
|
204
|
+
- Parse text results from MCP tool response
|
|
205
|
+
- Reduce prompt length on timeout
|
|
206
206
|
|
|
207
|
-
### Step 4:
|
|
207
|
+
### Step 4: Receive and Parse Codex Feedback
|
|
208
208
|
|
|
209
|
-
Codex
|
|
210
|
-
- Strengths:
|
|
211
|
-
- Issues:
|
|
212
|
-
- Open Questions:
|
|
213
|
-
- Summary:
|
|
209
|
+
Organize Codex response into this structure:
|
|
210
|
+
- Strengths: List of valid points
|
|
211
|
+
- Issues: Table of items needing improvement
|
|
212
|
+
- Open Questions: List of open questions
|
|
213
|
+
- Summary: Overall assessment
|
|
214
214
|
|
|
215
215
|
### Step 5: Claude Cross-check
|
|
216
216
|
|
|
217
|
-
**
|
|
217
|
+
**Verify items Codex may have missed:**
|
|
218
218
|
|
|
219
|
-
1.
|
|
219
|
+
1. **Check for already-resolved issues**
|
|
220
220
|
```bash
|
|
221
221
|
git log --oneline -20
|
|
222
222
|
```
|
|
223
|
-
- PRD
|
|
223
|
+
- Verify if issues mentioned in PRD were already resolved via commits
|
|
224
224
|
|
|
225
|
-
2.
|
|
226
|
-
- npm: `npm view [
|
|
227
|
-
- pip: `pip show [
|
|
228
|
-
-
|
|
225
|
+
2. **Validate package/dependency existence**
|
|
226
|
+
- npm: `npm view [package-name]`
|
|
227
|
+
- pip: `pip show [package-name]` or PyPI search
|
|
228
|
+
- Check for mentions of non-existent packages
|
|
229
229
|
|
|
230
|
-
3.
|
|
231
|
-
-
|
|
232
|
-
-
|
|
230
|
+
3. **Verify codebase-PRD synchronization**
|
|
231
|
+
- Confirm files/modules mentioned in PRD actually exist
|
|
232
|
+
- Check alignment between existing implementation and PRD design
|
|
233
233
|
|
|
234
|
-
4. **CLAUDE.md
|
|
235
|
-
-
|
|
236
|
-
- TaskMaster
|
|
234
|
+
4. **Verify CLAUDE.md guideline compliance**
|
|
235
|
+
- Project convention adherence
|
|
236
|
+
- TaskMaster workflow compatibility
|
|
237
237
|
|
|
238
|
-
5. **Codex
|
|
239
|
-
-
|
|
240
|
-
-
|
|
238
|
+
5. **Identify Codex errors**
|
|
239
|
+
- List incorrect items from validation
|
|
240
|
+
- Prepare evidence (git commit, actual files, etc.)
|
|
241
241
|
|
|
242
|
-
### Step 6:
|
|
242
|
+
### Step 6: Disagreement Check and Re-review (Context Re-transmission)
|
|
243
243
|
|
|
244
|
-
|
|
244
|
+
**If there are disagreements:**
|
|
245
245
|
|
|
246
|
-
`mcp__codex__codex`
|
|
246
|
+
Call `mcp__codex__codex` as new session, **including previous conversation context in prompt**:
|
|
247
247
|
|
|
248
|
-
####
|
|
248
|
+
#### Re-review Prompt Template:
|
|
249
249
|
```
|
|
250
|
-
##
|
|
251
|
-
|
|
250
|
+
## Role
|
|
251
|
+
You are a PRD review expert.
|
|
252
252
|
|
|
253
|
-
##
|
|
253
|
+
## Previous Review Context
|
|
254
254
|
|
|
255
|
-
###
|
|
256
|
-
[
|
|
255
|
+
### First Review Summary
|
|
256
|
+
[Key content from Codex first response - Strengths, Issues, Questions, Summary]
|
|
257
257
|
|
|
258
|
-
### Claude Cross-check
|
|
259
|
-
[
|
|
258
|
+
### Claude Cross-check Results
|
|
259
|
+
[Disagreement items and evidence]
|
|
260
260
|
|
|
261
|
-
##
|
|
262
|
-
|
|
263
|
-
1. [
|
|
264
|
-
2. [
|
|
261
|
+
## Re-review Request
|
|
262
|
+
Please re-review only the following items reflecting the cross-check results:
|
|
263
|
+
1. [Disagreement item 1]
|
|
264
|
+
2. [Disagreement item 2]
|
|
265
265
|
|
|
266
|
-
|
|
266
|
+
Respond only with modified sections from the original assessment.
|
|
267
267
|
|
|
268
|
-
##
|
|
269
|
-
[
|
|
268
|
+
## Reference Information
|
|
269
|
+
[Additional context if needed - related code, git log, etc.]
|
|
270
270
|
```
|
|
271
271
|
|
|
272
|
-
**MCP
|
|
272
|
+
**MCP Call:**
|
|
273
273
|
```
|
|
274
|
-
mcp__codex__codex
|
|
275
|
-
- prompt: [
|
|
274
|
+
mcp__codex__codex tool call:
|
|
275
|
+
- prompt: [re-review prompt]
|
|
276
276
|
- sandbox: "read-only"
|
|
277
277
|
```
|
|
278
278
|
|
|
279
|
-
|
|
280
|
-
-
|
|
281
|
-
-
|
|
282
|
-
- Codex
|
|
279
|
+
**Ping-pong termination conditions:**
|
|
280
|
+
- Consensus reached (no disagreement items)
|
|
281
|
+
- Maximum 3 iterations reached
|
|
282
|
+
- Codex accepts Claude's evidence
|
|
283
283
|
|
|
284
|
-
|
|
285
|
-
- 1
|
|
286
|
-
- 2
|
|
287
|
-
- 3
|
|
284
|
+
**Iteration tracking:**
|
|
285
|
+
- Round 1: Initial review
|
|
286
|
+
- Round 2: First re-review (includes previous context)
|
|
287
|
+
- Round 3: Second re-review (final)
|
|
288
288
|
|
|
289
|
-
### Step 7:
|
|
289
|
+
### Step 7: Reach Consensus and Deliver Results
|
|
290
290
|
|
|
291
|
-
|
|
291
|
+
**Output Format:**
|
|
292
292
|
|
|
293
293
|
```markdown
|
|
294
|
-
## PRD
|
|
294
|
+
## PRD Review Results (Codex + Claude Consensus)
|
|
295
295
|
|
|
296
|
-
###
|
|
297
|
-
-
|
|
298
|
-
-
|
|
296
|
+
### Review Process
|
|
297
|
+
- Ping-pong iterations: [N]
|
|
298
|
+
- Consensus status: [Full consensus / Partial consensus / Claude determination]
|
|
299
299
|
|
|
300
|
-
### [VALID]
|
|
301
|
-
|
|
|
302
|
-
|
|
300
|
+
### [VALID] Valid Feedback
|
|
301
|
+
| Item | Description | Source |
|
|
302
|
+
|------|-------------|--------|
|
|
303
303
|
|
|
304
|
-
### [ISSUE]
|
|
305
|
-
|
|
|
306
|
-
|
|
304
|
+
### [ISSUE] Items Needing Improvement
|
|
305
|
+
| Item | Problem | Recommended Fix | Source |
|
|
306
|
+
|------|---------|-----------------|--------|
|
|
307
307
|
|
|
308
|
-
### [CORRECTION] Codex
|
|
309
|
-
| Codex
|
|
310
|
-
|
|
308
|
+
### [CORRECTION] Codex Error Corrections
|
|
309
|
+
| Codex Claim | Actual Situation | Evidence |
|
|
310
|
+
|-------------|------------------|----------|
|
|
311
311
|
|
|
312
|
-
### [DECISION]
|
|
313
|
-
(
|
|
312
|
+
### [DECISION] Items Requiring Decision
|
|
313
|
+
(If there are options, ask via AskUserQuestion)
|
|
314
314
|
|
|
315
|
-
### [SUMMARY]
|
|
316
|
-
[
|
|
315
|
+
### [SUMMARY] Final Conclusion
|
|
316
|
+
[Final summary]
|
|
317
317
|
```
|
|
318
318
|
|
|
319
319
|
---
|
|
320
320
|
|
|
321
|
-
|
|
322
|
-
|
|
323
|
-
- 한글로 답변
|
|
324
|
-
- 코드나 문서 작성 시 이모지 사용 금지
|
|
325
|
-
- **불명확한 사항은 추측하지 말고 AskUserQuestion으로 질문**
|
|
326
|
-
- PRD 수정 제안은 사용자 확인 후 진행
|
|
327
|
-
- Codex와 Claude 의견 불일치 시 근거를 명확히 제시
|
|
328
|
-
- 검토 기준이 애매하면 사용자에게 우선순위 질문
|
|
321
|
+
> See [Work Guidelines](../guidelines/work-guidelines.md)
|
|
329
322
|
|
|
330
323
|
---
|
|
331
324
|
|
|
332
|
-
##
|
|
325
|
+
## Error Handling
|
|
333
326
|
|
|
334
|
-
- **PRD
|
|
335
|
-
- **Codex MCP
|
|
336
|
-
-
|
|
337
|
-
-
|
|
327
|
+
- **PRD file not found**: "File not found. Please verify the path."
|
|
328
|
+
- **Codex MCP call failed**: "Codex MCP tool call failed. Please check MCP server status."
|
|
329
|
+
- **Timeout**: "Codex response timeout. Please reduce prompt length or try again."
|
|
330
|
+
- **Re-review needed**: "Proceeding with re-review in new session including previous context."
|
|
338
331
|
|
|
339
332
|
---
|
|
340
333
|
|
|
341
|
-
## Codex MCP
|
|
334
|
+
## Codex MCP Tool Reference
|
|
342
335
|
|
|
343
|
-
###
|
|
344
|
-
|
|
|
345
|
-
|
|
346
|
-
| `mcp__codex__codex` | `prompt`, `sandbox`, `cwd`, `model
|
|
336
|
+
### MCP Tools Used by Claude
|
|
337
|
+
| Tool | Parameters | Description |
|
|
338
|
+
|------|------------|-------------|
|
|
339
|
+
| `mcp__codex__codex` | `prompt`, `sandbox`, `cwd`, `model`, etc. | Start new Codex session |
|
|
347
340
|
|
|
348
|
-
|
|
349
|
-
- `prompt` (
|
|
350
|
-
- `sandbox`: "read-only" (
|
|
351
|
-
- `cwd`:
|
|
352
|
-
- `model`:
|
|
341
|
+
**Key Parameters:**
|
|
342
|
+
- `prompt` (required): Initial prompt
|
|
343
|
+
- `sandbox`: "read-only" (only file reading allowed, safe)
|
|
344
|
+
- `cwd`: Working directory (optional)
|
|
345
|
+
- `model`: Model specification (optional, e.g., "o3", "o4-mini")
|
|
353
346
|
|
|
354
|
-
|
|
355
|
-
-
|
|
356
|
-
- CLI
|
|
347
|
+
**Notes:**
|
|
348
|
+
- For ping-pong, call new session with previous context included in prompt
|
|
349
|
+
- Unlike CLI's `codex resume`, MCP uses context re-transmission method
|
|
357
350
|
|
|
358
351
|
---
|
|
359
352
|
|
|
360
|
-
## Cross-check
|
|
353
|
+
## Cross-check Checklist
|
|
361
354
|
|
|
362
|
-
|
|
|
363
|
-
|
|
364
|
-
|
|
|
365
|
-
|
|
|
366
|
-
|
|
|
367
|
-
|
|
|
355
|
+
| Verification Item | Method | Example |
|
|
356
|
+
|-------------------|--------|---------|
|
|
357
|
+
| Already resolved issues | `git log --grep="issue-number"` | Specific issue already resolved via commit |
|
|
358
|
+
| Package existence | `npm view` / `pip show` | Non-existent SDK mentioned |
|
|
359
|
+
| File/module existence | `ls`, `find`, `grep` | Specific adapter file location |
|
|
360
|
+
| Version match | `package.json`, `requirements.txt` | Specified version vs actual version |
|
|
368
361
|
|
|
369
362
|
---
|
|
370
363
|
|
|
371
|
-
##
|
|
364
|
+
## Usage Examples
|
|
372
365
|
|
|
373
366
|
```bash
|
|
374
|
-
#
|
|
367
|
+
# Review after PRD generation
|
|
375
368
|
/tm:convert-prd .taskmaster/docs/my-idea.md
|
|
376
|
-
# prd.md
|
|
369
|
+
# prd.md generated
|
|
377
370
|
|
|
378
371
|
/tm:review-prd-with-codex .taskmaster/docs/prd.md
|
|
379
|
-
# Codex MCP
|
|
372
|
+
# Codex MCP review + Claude cross-check + ping-pong + consensus results output
|
|
380
373
|
```
|
|
381
374
|
|
|
382
375
|
---
|
|
383
376
|
|
|
384
|
-
##
|
|
377
|
+
## Workflow Diagram
|
|
385
378
|
|
|
386
379
|
```
|
|
387
|
-
[Step 1:
|
|
380
|
+
[Step 1: Gather Rich Context]
|
|
388
381
|
|
|
|
389
382
|
v
|
|
390
|
-
[Step 2:
|
|
383
|
+
[Step 2: Construct Prompt]
|
|
391
384
|
|
|
|
392
385
|
v
|
|
393
|
-
[Step 3: Codex MCP
|
|
394
|
-
| (mcp__codex__codex)
|
|
395
|
-
v
|
|
396
|
-
[Step 4:
|
|
397
|
-
|
|
|
398
|
-
v
|
|
399
|
-
[Step 5: Claude Cross-check]
|
|
400
|
-
|
|
|
401
|
-
v
|
|
402
|
-
[Step 6:
|
|
403
|
-
|
|
|
404
|
-
├─ YES &
|
|
405
|
-
|
|
|
386
|
+
[Step 3: Codex MCP First Review] ──────────────────────────┐
|
|
387
|
+
| (mcp__codex__codex) |
|
|
388
|
+
v |
|
|
389
|
+
[Step 4: Parse Feedback] |
|
|
390
|
+
| |
|
|
391
|
+
v |
|
|
392
|
+
[Step 5: Claude Cross-check] |
|
|
393
|
+
| |
|
|
394
|
+
v |
|
|
395
|
+
[Step 6: Disagreements?] |
|
|
396
|
+
| |
|
|
397
|
+
├─ YES & iterations < 3 ─> [New MCP call + context]────┘
|
|
398
|
+
| (includes conversation summary)
|
|
406
399
|
|
|
|
407
|
-
└─ NO
|
|
400
|
+
└─ NO or iterations >= 3 ─> [Step 7: Deliver Consensus]
|
|
408
401
|
```
|
|
409
402
|
|
|
410
|
-
|
|
411
|
-
1. Codex MCP
|
|
412
|
-
2. Claude
|
|
413
|
-
3.
|
|
414
|
-
4.
|
|
415
|
-
5.
|
|
403
|
+
**Core Flow:**
|
|
404
|
+
1. Execute Codex MCP review
|
|
405
|
+
2. Claude validates via cross-check
|
|
406
|
+
3. On disagreement, new MCP call (include previous context in prompt)
|
|
407
|
+
4. Repeat until consensus or maximum 3 iterations
|
|
408
|
+
5. Deliver final results
|
|
416
409
|
|
|
417
410
|
---
|
|
418
411
|
|
|
419
|
-
##
|
|
412
|
+
## Expected Output Example
|
|
420
413
|
|
|
421
414
|
```markdown
|
|
422
|
-
## PRD
|
|
423
|
-
|
|
424
|
-
###
|
|
425
|
-
-
|
|
426
|
-
-
|
|
427
|
-
|
|
428
|
-
### [VALID]
|
|
429
|
-
|
|
|
430
|
-
|
|
431
|
-
|
|
|
432
|
-
|
|
|
433
|
-
|
|
|
434
|
-
|
|
435
|
-
### [ISSUE]
|
|
436
|
-
|
|
|
437
|
-
|
|
438
|
-
|
|
|
439
|
-
| Milestone
|
|
440
|
-
|
|
441
|
-
### [CORRECTION] Codex
|
|
442
|
-
| Codex
|
|
443
|
-
|
|
444
|
-
|
|
|
445
|
-
|
|
446
|
-
### [SUMMARY]
|
|
447
|
-
PRD
|
|
448
|
-
|
|
415
|
+
## PRD Review Results (Codex + Claude Consensus)
|
|
416
|
+
|
|
417
|
+
### Review Process
|
|
418
|
+
- Ping-pong iterations: 2
|
|
419
|
+
- Consensus status: Full consensus
|
|
420
|
+
|
|
421
|
+
### [VALID] Valid Feedback
|
|
422
|
+
| Item | Description | Source |
|
|
423
|
+
|------|-------------|--------|
|
|
424
|
+
| No success metrics | Goals only list functional goals, no performance/cost criteria defined | Codex |
|
|
425
|
+
| Missing error handling | No exception handling in API call code | Codex + Claude |
|
|
426
|
+
| Test scenarios | Only happy-path exists, no failure cases | Codex |
|
|
427
|
+
|
|
428
|
+
### [ISSUE] Items Needing Improvement
|
|
429
|
+
| Item | Problem | Recommended Fix | Source |
|
|
430
|
+
|------|---------|-----------------|--------|
|
|
431
|
+
| Dependency list | Non-existent package mentioned | Update to actual package | Claude verification |
|
|
432
|
+
| Milestone status | Includes already-resolved issues | Mark complete or remove | Claude verification |
|
|
433
|
+
|
|
434
|
+
### [CORRECTION] Codex Error Corrections
|
|
435
|
+
| Codex Claim | Actual Situation | Evidence |
|
|
436
|
+
|-------------|------------------|----------|
|
|
437
|
+
| Feature not implemented | Feature already exists | Verified in src/modules/ folder |
|
|
438
|
+
|
|
439
|
+
### [SUMMARY] Final Conclusion
|
|
440
|
+
PRD is structurally sound, but non-functional requirements (performance, security, error handling)
|
|
441
|
+
and dependency information need updates. Recommend removing already-resolved issues from Milestone.
|
|
449
442
|
```
|