ctx-cc 2.3.0 → 3.1.0

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/README.md CHANGED
@@ -1,6 +1,25 @@
1
- # CTX 2.3 - Continuous Task eXecution
1
+ <div align="center">
2
2
 
3
- > Smart workflow orchestration for Claude Code. PRD-driven. Design-first. 8 commands. Debug loop until 100% fixed.
3
+ # CTX
4
+
5
+ ### Continuous Task eXecution
6
+
7
+ **Intelligent workflow orchestration for Claude Code.**
8
+
9
+ [![npm version](https://img.shields.io/npm/v/ctx-cc.svg?style=flat-square)](https://www.npmjs.com/package/ctx-cc)
10
+ [![npm downloads](https://img.shields.io/npm/dm/ctx-cc.svg?style=flat-square)](https://www.npmjs.com/package/ctx-cc)
11
+ [![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg?style=flat-square)](https://opensource.org/licenses/MIT)
12
+ [![GitHub stars](https://img.shields.io/github/stars/jufjuf/CTX?style=flat-square)](https://github.com/jufjuf/CTX/stargazers)
13
+
14
+ <img src="./assets/terminal.png" alt="CTX Terminal" width="700">
15
+
16
+ **Task parallelization. Pre-commit review. Criteria auto-generation. Smart handoff. Issue tracker sync.**
17
+
18
+ [Installation](#installation) · [Quick Start](#quick-start) · [New in 3.1](#new-in-31) · [Commands](#commands) · [Why CTX](#why-ctx)
19
+
20
+ </div>
21
+
22
+ ---
4
23
 
5
24
  ## Installation
6
25
 
@@ -8,97 +27,230 @@
8
27
  npx ctx-cc
9
28
  ```
10
29
 
11
- Options:
30
+ That's it. CTX installs itself to your Claude Code environment.
31
+
12
32
  ```bash
33
+ # Options
13
34
  npx ctx-cc --global # Install to ~/.claude (default)
14
35
  npx ctx-cc --project # Install to .claude in current directory
15
36
  npx ctx-cc --force # Overwrite existing installation
16
37
  ```
17
38
 
18
- ## Why CTX 2.3?
19
-
20
- | Feature | Before | CTX 2.3 |
21
- |---------|--------|---------|
22
- | Requirements | Ad-hoc goals | **PRD.json with stories** |
23
- | Design | Manual workflow | **Full design system** |
24
- | Accessibility | Manual checks | **WCAG 2.2 AA + EAA 2025** |
25
- | Tokens | Custom formats | **W3C Design Tokens** |
26
- | Verification | Task-based | **Acceptance criteria** |
27
- | Commands | 12-27 | **8** (organized) |
28
- | Router | Manual | **Smart (auto-routing)** |
29
- | Debug | Manual | **Loop until 100% fixed** |
30
- | Browser Verify | No | **Playwright/DevTools** |
31
- | Planning | Any size | **Atomic (2-3 tasks max)** |
32
- | Resume cost | ~50k tokens | **~2.5k tokens** |
39
+ ---
33
40
 
34
41
  ## Quick Start
35
42
 
36
43
  ```
37
- 1. /ctx init Gather ALL info upfront (requirements + credentials)
38
- 2. /ctx Autonomous execution with minimal interruption
39
- 3. /ctx pause Checkpoint when needed
44
+ 1. /ctx init Gather requirements + credentials + design context
45
+ 2. /ctx map Build repository map (existing codebases)
46
+ 3. /ctx Autonomous execution with minimal interruption
40
47
  ```
41
48
 
42
49
  **The Flow:**
43
50
  ```
44
- /ctx init → Gather everything → /ctx → Autonomous loop → Delivered!
51
+ /ctx init → /ctx map → /ctx → Autonomous loop → Delivered!
52
+ ```
53
+
54
+ ---
55
+
56
+ ## New in 3.1
57
+
58
+ ### Intelligent Task Parallelization
59
+ Tasks without dependencies run simultaneously:
60
+ ```
61
+ Wave 1: [T001, T003] → Parallel (no deps)
62
+ Wave 2: [T002] → After T001
63
+ Wave 3: [T004] → After T002
64
+
65
+ Result: 40% faster execution
66
+ ```
67
+
68
+ ### Pre-Commit Review (ctx-reviewer)
69
+ Catches errors BEFORE they're committed:
70
+ - Type errors (TypeScript, Python, Go)
71
+ - Unresolved imports
72
+ - Circular dependencies
73
+ - Security vulnerabilities
74
+ - Empty catch blocks, console.logs
75
+
76
+ ```
77
+ [CTX] Pre-Commit Review
78
+ ✅ Types: Pass
79
+ ✅ Imports: Pass
80
+ ⚠️ Medium: 2 console.log statements
81
+ ❌ Critical: SQL injection risk at line 45
82
+
83
+ Status: BLOCKED - Fix critical issue
84
+ ```
85
+
86
+ ### Acceptance Criteria Auto-Generation
87
+ AI suggests comprehensive criteria:
88
+ ```
89
+ Story: "Add user authentication"
90
+
91
+ Suggested Criteria:
92
+ ✓ User can register with email/password
93
+ ✓ Invalid credentials show error
94
+ ✓ Passwords hashed with bcrypt
95
+ ✓ Session expires after 24h
96
+ ✓ Brute force protection enabled
97
+
98
+ [A] Accept all [B] See more [C] Edit
99
+ ```
100
+
101
+ ### Smart Context Handoff
102
+ Seamless transitions at context limits:
103
+
104
+ | Threshold | Action |
105
+ |-----------|--------|
106
+ | 40% | Prepare handoff notes |
107
+ | 50% | Write HANDOFF.md, warn |
108
+ | 60% | Spawn fresh agent |
109
+
110
+ Zero information loss. Work continues automatically.
111
+
112
+ ### Issue Tracker Integration
113
+ Sync with Linear, Jira, or GitHub Issues:
114
+ ```bash
115
+ /ctx integrate linear # Setup Linear
116
+ /ctx integrate jira # Setup Jira
117
+ /ctx integrate github # Setup GitHub Issues
118
+ /ctx integrate --sync # Force sync all stories
119
+ ```
120
+
121
+ Features:
122
+ - Bidirectional story sync
123
+ - Status mapping (CTX → tracker)
124
+ - Auto-close on verify pass
125
+ - Comment on verify fail
126
+
127
+ ---
128
+
129
+ ## From 3.0
130
+
131
+ ### Repository Mapping (like Aider)
132
+ ```bash
133
+ /ctx map # Build token-optimized codebase map
134
+ /ctx map --expand # Include call graph (8k tokens)
135
+ /ctx map --refresh # Force full rebuild
45
136
  ```
46
137
 
47
- ## Front-Loaded Approach
138
+ Creates `REPO-MAP.md` with symbols, dependencies, and navigation hints.
48
139
 
49
- CTX gathers EVERYTHING at initialization:
50
- - **Requirements** → PRD.json stories
51
- - **Acceptance criteria** How to verify each story
52
- - **Test credentials** .ctx/.env (gitignored)
53
- - **Constitution** → Rules for autonomous decisions
54
- - **Design context** → Brand personality, inspirations, accessibility needs
140
+ ### Discussion Phase (like GSD)
141
+ ```bash
142
+ /ctx discuss S001 # Capture decisions BEFORE planning
143
+ /ctx discuss --review # Review locked decisions
144
+ ```
55
145
 
56
- Then executes autonomously:
57
- - Only interrupts for architecture decisions (Rule 4)
58
- - Only interrupts for design approvals at gates
59
- - Uses stored credentials for browser testing
60
- - Loops through stories until all pass
146
+ Prevents mid-implementation questions by locking decisions in `CONTEXT.md`.
61
147
 
62
- ## Design System (New in 2.3)
148
+ ### Model Profiles (Cost Optimization)
149
+ ```bash
150
+ /ctx profile # Show current profile
151
+ /ctx profile quality # Best models (Opus everywhere)
152
+ /ctx profile balanced # Smart mix (default)
153
+ /ctx profile budget # Fast models (60% savings)
154
+ ```
63
155
 
64
- CTX handles visual work with dedicated agents and approval gates.
156
+ | Profile | Research | Execute | Verify | Cost |
157
+ |---------|----------|---------|--------|------|
158
+ | quality | Opus | Opus | Sonnet | 3x |
159
+ | balanced | Opus | Sonnet | Haiku | 1x |
160
+ | budget | Sonnet | Sonnet | Haiku | 0.4x |
65
161
 
66
- ### Story Types
67
- | Type | Agent | Purpose |
68
- |------|-------|---------|
69
- | feature | ctx-executor | Standard implementation |
70
- | brand | ctx-designer | BRAND_KIT.md + W3C tokens |
71
- | design | ctx-designer | UI components/pages |
162
+ ### Git-Native Workflow
163
+ Every completed task auto-commits:
164
+ ```
165
+ [CTX] Implement user login endpoint
72
166
 
73
- ### 3 Options Pattern
74
- All design decisions present three options:
75
- - **Option A**: Conservative (safe, proven)
76
- - **Option B**: Balanced (recommended)
77
- - **Option C**: Bold (distinctive)
167
+ Story: S001 - User Authentication
168
+ Criteria: User can log in with credentials
169
+ Files: src/auth/login.ts, src/routes/auth.ts
78
170
 
79
- ### Design Approval Gates
171
+ Co-Authored-By: Claude <noreply@anthropic.com>
80
172
  ```
81
- Mood Board → Direction (A/B/C) → Prototype → Final
173
+
174
+ Configure in `.ctx/config.json`:
175
+ ```json
176
+ {
177
+ "git": {
178
+ "autoCommit": true,
179
+ "commitPerTask": true
180
+ }
181
+ }
82
182
  ```
83
- Each gate requires user approval before proceeding.
84
183
 
85
- ### WCAG 2.2 AA + EAA 2025
86
- | Criterion | Requirement |
87
- |-----------|-------------|
88
- | 2.4.11 | Focus not obscured |
89
- | 2.5.7 | Drag alternatives |
90
- | 2.5.8 | 24x24px targets |
91
- | 3.3.8 | Accessible auth |
184
+ ### Persistent Debug State
185
+ Debug sessions survive context resets:
186
+ ```bash
187
+ /ctx debug --resume # Continue previous session
188
+ ```
92
189
 
93
- EU markets require EAA 2025 compliance (enforcement: June 28, 2025).
190
+ State stored in `.ctx/debug/sessions/`:
191
+ - `STATE.json` - Machine-readable progress
192
+ - `TRACE.md` - Human-readable log
193
+ - `hypotheses.json` - All theories tested
194
+ - `screenshots/` - Visual evidence
94
195
 
95
- ## The 8 Commands
196
+ ### Parallel Codebase Analysis
197
+ ```bash
198
+ /ctx map-codebase # Full analysis with 4 parallel agents
199
+ ```
200
+
201
+ Spawns 4 agents simultaneously:
202
+ | Agent | Output | Analyzes |
203
+ |-------|--------|----------|
204
+ | TECH | TECH.md | Languages, frameworks, dependencies |
205
+ | ARCH | ARCH.md | Patterns, data flow, modules |
206
+ | QUALITY | QUALITY.md | Test coverage, lint, type safety |
207
+ | CONCERNS | CONCERNS.md | Security, tech debt, performance |
208
+
209
+ Results synthesized into `SUMMARY.md`.
210
+
211
+ ---
212
+
213
+ ## Why CTX?
214
+
215
+ | Feature | Aider | GSD | CTX 3.0 |
216
+ |---------|-------|-----|---------|
217
+ | Repository Map | Yes | No | **Yes** |
218
+ | Discussion Phase | No | Yes | **Yes** |
219
+ | Model Profiles | Yes | Partial | **Yes** |
220
+ | Git-Native Commits | Yes | No | **Yes** |
221
+ | Persistent Debug | No | Partial | **Yes** |
222
+ | Parallel Analysis | No | Yes | **Yes** |
223
+ | PRD-Driven | No | Yes | **Yes** |
224
+ | Design System | No | No | **Yes** |
225
+ | Browser Verification | No | No | **Yes** |
226
+
227
+ **CTX 3.0 combines the best of Aider and GSD.**
228
+
229
+ ---
230
+
231
+ ## Commands
96
232
 
97
233
  ### Smart (Auto-routing)
98
234
  | Command | Purpose |
99
235
  |---------|---------|
100
236
  | `/ctx` | **Smart router** - reads STATE.md, does the right thing |
101
- | `/ctx init` | Initialize project with STATE.md |
237
+ | `/ctx init` | Initialize project with STATE.md + PRD.json |
238
+
239
+ ### Mapping
240
+ | Command | Purpose |
241
+ |---------|---------|
242
+ | `/ctx map` | Build repository map (REPO-MAP.md) |
243
+ | `/ctx map-codebase` | Deep analysis (4 parallel agents) |
244
+
245
+ ### Discussion
246
+ | Command | Purpose |
247
+ |---------|---------|
248
+ | `/ctx discuss [story]` | Capture decisions before planning |
249
+
250
+ ### Configuration
251
+ | Command | Purpose |
252
+ |---------|---------|
253
+ | `/ctx profile [name]` | Switch model profile (quality/balanced/budget) |
102
254
 
103
255
  ### Inspect (Read-only)
104
256
  | Command | Purpose |
@@ -123,206 +275,197 @@ EU markets require EAA 2025 compliance (enforcement: June 28, 2025).
123
275
  | `/ctx phase list` | Show all phases |
124
276
  | `/ctx phase add "goal"` | Add new phase |
125
277
  | `/ctx phase next` | Complete current, move to next |
126
- | `/ctx phase skip` | Skip current phase |
127
-
128
- ### Smart Router States
129
278
 
130
- | State | What `/ctx` does |
131
- |-------|------------------|
132
- | initializing | Research + Plan (ArguSeek + ChunkHound) |
133
- | executing | Execute current task |
134
- | debugging | **Debug loop until 100% fixed** |
135
- | verifying | Three-level verification |
136
- | paused | Resume from checkpoint |
279
+ ### Integration
280
+ | Command | Purpose |
281
+ |---------|---------|
282
+ | `/ctx integrate` | Show integration status |
283
+ | `/ctx integrate linear` | Setup Linear |
284
+ | `/ctx integrate jira` | Setup Jira |
285
+ | `/ctx integrate github` | Setup GitHub Issues |
286
+ | `/ctx integrate --sync` | Sync all stories |
137
287
 
138
- ## Debug Loop (Key Feature)
288
+ ---
139
289
 
140
- When something breaks, CTX enters debug mode and loops until fixed:
290
+ ## State Machine
141
291
 
142
292
  ```
143
- Loop (max 5 attempts):
144
- 1. Analyze error
145
- 2. Form hypothesis
146
- 3. Apply fix
147
- 4. Verify (build + tests + browser)
148
- 5. If fixed → done
149
- If not → new hypothesis, try again
293
+ initializing discussing → executing → verifying → COMPLETE
294
+ ↑ ↓
295
+ └── debugging ──┘
150
296
  ```
151
297
 
152
- **Browser verification for UI:**
153
- - Navigates to affected page
154
- - Checks elements exist
155
- - Takes screenshot proof
156
- - Saves to `.ctx/debug/`
157
-
158
- ## Key Design Principles
159
-
160
- ### Atomic Planning (2-3 Tasks Max)
161
- Why? Context degradation is real:
162
- | Context | Quality |
163
- |---------|---------|
164
- | 0-30% | Peak |
165
- | 30-50% | Good |
166
- | 50%+ | Degrading |
167
-
168
- Big work = multiple phases, not bigger plans.
169
-
170
- ### 95% Auto-Deviation Handling
171
- | Trigger | Action |
172
- |---------|--------|
173
- | Bug in existing code | Auto-fix |
174
- | Missing validation | Auto-add |
175
- | Blocking issue | Auto-fix |
176
- | Architecture decision | Ask user |
177
-
178
- ### Three-Level Verification
179
- 1. **Exists** - File on disk?
180
- 2. **Substantive** - Real code, not stub?
181
- 3. **Wired** - Imported and used?
182
-
183
- ### STATE.md - Single Source of Truth
184
- ~100 lines. Always accurate. Always read first.
185
-
186
- ## 6 Specialized Agents
187
-
188
- | Agent | Spawned when |
298
+ | State | What happens |
189
299
  |-------|--------------|
190
- | ctx-researcher | status = initializing |
191
- | ctx-planner | after research |
192
- | ctx-executor | status = executing (feature stories) |
193
- | ctx-designer | status = executing (brand/design stories) |
194
- | ctx-debugger | status = debugging |
195
- | ctx-verifier | status = verifying |
300
+ | initializing | Research + Map + Plan |
301
+ | discussing | Capture decisions in CONTEXT.md |
302
+ | executing | Execute with git-native commits |
303
+ | debugging | Persistent debug loop (max 10 attempts) |
304
+ | verifying | Three-level verification |
305
+ | paused | Resume from checkpoint |
196
306
 
197
- ## Integrations
307
+ ---
198
308
 
199
- ### ArguSeek (Web Research)
200
- Auto-runs during planning:
201
- - Best practices for the goal
202
- - Security considerations
203
- - Performance patterns
309
+ ## Context Management
204
310
 
205
- ### ChunkHound (Semantic Code Search)
206
- Auto-runs during planning:
207
- - Semantic search for relevant code
208
- - Pattern detection
209
- - Entry point mapping
311
+ CTX actively manages context budget:
210
312
 
211
- Install: `uv tool install chunkhound`
313
+ | Usage | Quality | Action |
314
+ |-------|---------|--------|
315
+ | 0-30% | Peak | Continue |
316
+ | 30-40% | Good | Continue |
317
+ | 40-50% | Good | Prepare handoff notes |
318
+ | 50-60% | Degrading | Auto-checkpoint |
319
+ | 60-70% | Degrading | Create HANDOFF.md |
320
+ | 70%+ | Poor | Force checkpoint |
212
321
 
213
- ### Browser Verification (Playwright/Chrome DevTools)
214
- Auto-runs during debugging and verification:
215
- - Navigate to pages
216
- - Check elements exist
217
- - Take screenshot proof
322
+ Smart handoff creates `HANDOFF.md` with:
323
+ - Completed tasks with commit hashes
324
+ - Current task progress
325
+ - Key decisions made
326
+ - Files modified
327
+ - Next steps
218
328
 
219
- ### Figma MCP (Design Context)
220
- Auto-runs during design stories:
221
- - Extract design tokens
222
- - Get component metadata
223
- - Screenshot references
329
+ ---
224
330
 
225
- ### Gemini Design MCP (Visual Generation)
226
- Auto-runs during design stories:
227
- - Generate UI mockups
228
- - Create UI code
229
- - Analyze designs for accessibility
331
+ ## 14 Specialized Agents
332
+
333
+ | Agent | Spawned when | Model (balanced) |
334
+ |-------|--------------|------------------|
335
+ | ctx-mapper | /ctx map | haiku |
336
+ | ctx-tech-mapper | /ctx map-codebase | haiku |
337
+ | ctx-arch-mapper | /ctx map-codebase | haiku |
338
+ | ctx-quality-mapper | /ctx map-codebase | haiku |
339
+ | ctx-concerns-mapper | /ctx map-codebase | haiku |
340
+ | ctx-discusser | status = discussing | sonnet |
341
+ | ctx-researcher | status = initializing | opus |
342
+ | ctx-planner | after research | opus |
343
+ | ctx-executor | status = executing | sonnet |
344
+ | ctx-designer | design stories | sonnet |
345
+ | ctx-debugger | status = debugging | sonnet |
346
+ | ctx-verifier | status = verifying | haiku |
347
+ | ctx-parallelizer | before execution | haiku |
348
+ | ctx-reviewer | before commit | sonnet |
349
+ | ctx-criteria-suggester | during init/discuss | sonnet |
350
+ | ctx-handoff | at context thresholds | haiku |
351
+
352
+ ---
230
353
 
231
354
  ## Directory Structure
232
355
 
233
356
  ```
234
357
  .ctx/
358
+ ├── config.json # Model profiles, git settings
235
359
  ├── STATE.md # Living digest - execution state
236
- ├── PRD.json # Requirements contract - stories + criteria
237
- ├── phases/{story_id}/ # Per-story data
238
- ├── RESEARCH.md # ArguSeek + ChunkHound results
239
- ├── PLAN.md # Tasks mapped to acceptance criteria
240
- ├── VERIFY.md # Verification report
241
- │ ├── MOOD_BOARD.md # Design references (design stories)
242
- └── DESIGN_BRIEF.md # Design decisions (design stories)
360
+ ├── PRD.json # Requirements contract
361
+ ├── REPO-MAP.md # Token-optimized codebase map
362
+ ├── REPO-MAP.json # Structured map data
363
+ ├── .env # Test credentials (GITIGNORED)
364
+ ├── codebase/ # Deep analysis results
365
+ │ ├── TECH.md
366
+ ├── ARCH.md
367
+ │ ├── QUALITY.md
368
+ │ ├── CONCERNS.md
369
+ │ └── SUMMARY.md
370
+ ├── phases/{story_id}/
371
+ │ ├── CONTEXT.md # Locked decisions (discussion phase)
372
+ │ ├── RESEARCH.md # ArguSeek results
373
+ │ ├── PLAN.md # Tasks mapped to criteria
374
+ │ └── VERIFY.md # Verification report
375
+ ├── debug/
376
+ │ ├── sessions/ # Persistent debug state
377
+ │ └── screenshots/ # Visual proof
243
378
  ├── checkpoints/ # Auto-checkpoints
244
- ├── debug/ # Debug screenshots
245
379
  └── memory/ # Decision memory
246
-
247
- project/
248
- ├── BRAND_KIT.md # Visual foundation (brand stories)
249
- └── tokens/ # W3C design tokens
250
- ├── primitive.tokens.json
251
- ├── semantic.tokens.json
252
- └── component.tokens.json
253
380
  ```
254
381
 
255
- ## PRD.json - Requirements Contract
382
+ ---
383
+
384
+ ## Configuration
256
385
 
386
+ `.ctx/config.json`:
257
387
  ```json
258
388
  {
259
- "brand": {
260
- "hasBrandKit": false,
261
- "personality": ["professional", "modern"],
262
- "euMarket": true
389
+ "activeProfile": "balanced",
390
+ "models": {
391
+ "architect": { "id": "claude-opus-4", "costTier": "high" },
392
+ "default": { "id": "claude-sonnet-4", "costTier": "medium" },
393
+ "fast": { "id": "claude-haiku-4", "costTier": "low" }
263
394
  },
264
- "design": {
265
- "wcagLevel": "AA",
266
- "eaaCompliance": true,
267
- "tokenFormat": "w3c"
268
- },
269
- "stories": [
270
- {
271
- "id": "S001",
272
- "type": "brand",
273
- "title": "Establish brand kit",
274
- "acceptanceCriteria": [
275
- "BRAND_KIT.md exists",
276
- "tokens/ directory populated"
277
- ],
278
- "passes": false
395
+ "profiles": {
396
+ "quality": {
397
+ "research": "architect",
398
+ "discussion": "architect",
399
+ "planning": "architect",
400
+ "execution": "architect"
401
+ },
402
+ "balanced": {
403
+ "research": "architect",
404
+ "discussion": "default",
405
+ "planning": "architect",
406
+ "execution": "default"
279
407
  },
280
- {
281
- "id": "S002",
282
- "type": "design",
283
- "title": "Login page",
284
- "acceptanceCriteria": [
285
- "WCAG 2.2 AA compliant",
286
- "All states implemented"
287
- ],
288
- "passes": false
408
+ "budget": {
409
+ "research": "default",
410
+ "planning": "default",
411
+ "execution": "default"
289
412
  }
290
- ],
291
- "metadata": {
292
- "currentStory": "S001",
293
- "passedStories": 0,
294
- "totalStories": 2
413
+ },
414
+ "git": {
415
+ "autoCommit": true,
416
+ "commitPerTask": true
295
417
  }
296
418
  }
297
419
  ```
298
420
 
299
- When a story passes verification, `passes` becomes `true`.
300
- When all stories pass, project is complete.
421
+ ---
301
422
 
302
- ## Secure Credentials (.ctx/.env)
423
+ ## Integrations
303
424
 
304
- During `/ctx init`, you'll be asked for test credentials:
425
+ ### ArguSeek (Web Research)
426
+ Auto-runs during planning for best practices, security, and patterns.
305
427
 
428
+ ### ChunkHound (Semantic Code Search)
429
+ Auto-runs during planning for semantic search and pattern detection.
306
430
  ```bash
307
- # .ctx/.env (automatically gitignored)
308
- APP_URL=http://localhost:3000
309
- TEST_USER_EMAIL=test@example.com
310
- TEST_USER_PASSWORD=testpass123
311
- ADMIN_EMAIL=admin@example.com
312
- ADMIN_PASSWORD=adminpass123
313
- API_KEY=your-api-key
431
+ uv tool install chunkhound
314
432
  ```
315
433
 
316
- **Why?**
317
- - Enables autonomous browser verification
318
- - No interruptions asking "what's the login?"
319
- - Agents use credentials silently for testing
320
- - NEVER echoed in logs or committed
434
+ ### Browser Verification (Playwright/Chrome DevTools)
435
+ Auto-runs during debugging and verification for visual proof.
436
+
437
+ ### Figma MCP (Design Context)
438
+ Auto-runs during design stories for tokens and component metadata.
439
+
440
+ ### Gemini Design MCP (Visual Generation)
441
+ Auto-runs during design stories for mockups and UI code.
442
+
443
+ ---
444
+
445
+ ## Key Principles
446
+
447
+ ### 95% Auto-Deviation Handling
448
+
449
+ | Trigger | Action |
450
+ |---------|--------|
451
+ | Bug in existing code | Auto-fix, document in commit |
452
+ | Missing validation | Auto-add, document |
453
+ | Blocking issue | Auto-fix, document |
454
+ | Architecture decision | **Ask user** |
455
+
456
+ ### Three-Level Verification
457
+
458
+ | Level | Question | Check |
459
+ |-------|----------|-------|
460
+ | Exists | File on disk? | Glob |
461
+ | Substantive | Real code, not stub? | No TODOs, no placeholders |
462
+ | Wired | Imported and used? | Trace imports |
321
463
 
322
- **Security:**
323
- - `.ctx/.gitignore` automatically protects `.env`
324
- - Credentials used ONLY for test automation
325
- - Never hardcoded, always read from .env
464
+ ### Atomic Planning
465
+
466
+ Plans limited to 2-3 tasks to prevent context degradation.
467
+
468
+ ---
326
469
 
327
470
  ## Updating
328
471
 
@@ -330,15 +473,18 @@ API_KEY=your-api-key
330
473
  npx ctx-cc --force
331
474
  ```
332
475
 
476
+ ---
477
+
333
478
  ## License
334
479
 
335
480
  MIT
336
481
 
337
- ## Links
482
+ ---
483
+
484
+ <div align="center">
338
485
 
339
- - [GitHub](https://github.com/jufjuf/CTX)
340
- - [Issues](https://github.com/jufjuf/CTX/issues)
486
+ **[GitHub](https://github.com/jufjuf/CTX)** · **[Issues](https://github.com/jufjuf/CTX/issues)** · **[npm](https://www.npmjs.com/package/ctx-cc)**
341
487
 
342
- ---
488
+ *CTX 3.1 - Task parallelization. Pre-commit review. Smart handoff. Issue tracker sync.*
343
489
 
344
- *CTX 2.3 - PRD-driven, design-first, story-verified, debug loop until 100% fixed*
490
+ </div>