ctx-cc 3.3.8 → 3.4.1

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.
@@ -1,7 +1,7 @@
1
1
  ---
2
2
  name: ctx-arch-mapper
3
3
  description: Architecture mapper for CTX 3.0. Analyzes patterns, data flow, modules, and entry points. Part of parallel codebase mapping.
4
- tools: Read, Bash, Glob, Grep
4
+ tools: Read, Write, Bash, Glob, Grep
5
5
  color: purple
6
6
  ---
7
7
 
@@ -1,7 +1,7 @@
1
1
  ---
2
2
  name: ctx-concerns-mapper
3
3
  description: Concerns mapper for CTX 3.0. Analyzes security vulnerabilities, tech debt, performance issues, and risks. Part of parallel codebase mapping.
4
- tools: Read, Bash, Glob, Grep
4
+ tools: Read, Write, Bash, Glob, Grep
5
5
  color: red
6
6
  ---
7
7
 
@@ -1,7 +1,7 @@
1
1
  ---
2
2
  name: ctx-quality-mapper
3
3
  description: Quality mapper for CTX 3.0. Analyzes test coverage, lint status, type safety, and code smells. Part of parallel codebase mapping.
4
- tools: Read, Bash, Glob, Grep
4
+ tools: Read, Write, Bash, Glob, Grep
5
5
  color: green
6
6
  ---
7
7
 
@@ -1,7 +1,7 @@
1
1
  ---
2
2
  name: ctx-tech-mapper
3
3
  description: Tech stack mapper for CTX 3.0. Analyzes languages, frameworks, dependencies, and versions. Part of parallel codebase mapping.
4
- tools: Read, Bash, Glob, Grep
4
+ tools: Read, Write, Bash, Glob, Grep
5
5
  color: blue
6
6
  ---
7
7
 
package/commands/ctx.md CHANGED
@@ -1,525 +1,78 @@
1
1
  ---
2
2
  name: ctx
3
- description: Smart router - understands natural language, detects intent, and guides users from first prompt. No commands to memorize.
3
+ description: Smart router - understands natural language, detects intent, and routes to the right workflow.
4
+ allowed-tools:
5
+ - Read
6
+ - Write
7
+ - Bash
8
+ - Glob
9
+ - Grep
10
+ - Task
11
+ - TaskOutput
12
+ - AskUserQuestion
13
+ - mcp__arguseek__research_iteratively
14
+ - mcp__arguseek__fetch_url
4
15
  ---
5
16
 
6
17
  <objective>
7
18
  CTX Smart Router - Conversational-first workflow orchestration.
8
19
 
9
- **Users don't need to know commands.** They just describe what they want, and CTX routes to the right workflow automatically.
20
+ **Users don't need to know commands.** They describe what they want, CTX routes automatically.
10
21
 
11
22
  ```
12
- User: "I want to build a todo app" Routes to init
13
- User: "Fix the login bug" Routes to debug
14
- User: "Is my app accessible?" Routes to QA
15
- User: "What should I do next?" Shows status + recommendation
23
+ User: "I want to build a todo app" -> Routes to init
24
+ User: "Fix the login bug" -> Routes to debug
25
+ User: "Study the codebase" -> Routes to map-codebase
26
+ User: "What should I do next?" -> Shows status + routes
16
27
  ```
17
28
  </objective>
18
29
 
19
- <workflow>
20
-
21
- ## Step 0: Intent Detection (ALWAYS RUNS FIRST)
22
-
23
- Before checking state, parse the user's natural language request.
24
-
25
- ### Intent Patterns
26
-
27
- | User Says | Detected Intent | Route To |
28
- |-----------|-----------------|----------|
29
- | "build", "create", "make", "start new", "I want to..." | `new-project` | /ctx:init |
30
- | "fix", "bug", "broken", "not working", "error" | `debug` | /ctx:debug |
31
- | "test", "QA", "check", "accessible", "works?" | `qa` | /ctx:qa |
32
- | "review", "audit", "security", "ready?" | `review` | /ctx:verify |
33
- | "production", "prod ready", "production ready" | `production-ready` | Full audit → QA → fixes |
34
- | "study", "analyze", "understand", "learn", "explore" | `analyze` | /ctx:map-codebase |
35
- | "improve", "optimize", "better", "enhance" | `improve` | Analyze → suggest → fix |
36
- | "deploy", "ship", "publish", "release" | `ship` | /ctx:verify → ship |
37
- | "status", "progress", "where", "what's next" | `status` | /ctx:status |
38
- | "help", "how", "what can", "commands" | `help` | /ctx:help |
39
- | "plan", "design", "architect" | `plan` | /ctx:plan |
40
- | "continue", "next", "go", "do it" | `continue` | (read STATE.md) |
41
-
42
- ### Intent Detection Logic
43
-
44
- ```javascript
45
- function detectIntent(userMessage) {
46
- const msg = userMessage.toLowerCase();
47
-
48
- // New project signals
49
- if (msg.match(/\b(build|create|make|start|new|want to)\b.*\b(app|project|feature|site|api)\b/)) {
50
- return 'new-project';
51
- }
52
-
53
- // Debug signals
54
- if (msg.match(/\b(fix|bug|broken|error|crash|not working|fails?|issue)\b/)) {
55
- return 'debug';
56
- }
57
-
58
- // QA signals
59
- if (msg.match(/\b(test|qa|check|accessible|works|validate|verify all)\b/)) {
60
- return 'qa';
61
- }
62
-
63
- // Production-ready signals
64
- if (msg.match(/\b(production|prod ready|production.?ready|make it ready)\b/)) {
65
- return 'production-ready';
66
- }
67
-
68
- // Analyze/study signals
69
- if (msg.match(/\b(study|analyze|understand|learn|explore|what is this)\b/)) {
70
- return 'analyze';
71
- }
72
-
73
- // Improve signals
74
- if (msg.match(/\b(improve|optimize|better|enhance|upgrade|refactor)\b/)) {
75
- return 'improve';
76
- }
77
-
78
- // Review signals
79
- if (msg.match(/\b(review|audit|security|ready|before deploy)\b/)) {
80
- return 'review';
81
- }
82
-
83
- // Ship signals
84
- if (msg.match(/\b(deploy|ship|publish|release|push|launch)\b/)) {
85
- return 'ship';
86
- }
87
-
88
- // Status signals
89
- if (msg.match(/\b(status|progress|where|what.?s next|current)\b/)) {
90
- return 'status';
91
- }
92
-
93
- // Help signals
94
- if (msg.match(/\b(help|how|what can|commands|guide)\b/)) {
95
- return 'help';
96
- }
97
-
98
- // Continue signals (or no clear intent)
99
- return 'continue';
100
- }
101
- ```
102
-
103
- ## Step 1: Check Project State
104
-
105
- Check if `.ctx/` folder exists:
106
-
107
- ### If NO `.ctx/` folder (New User)
108
-
109
- Based on detected intent:
110
-
111
- | Intent | Action |
112
- |--------|--------|
113
- | `new-project` | "Let's set up your project!" → Run /ctx:init flow |
114
- | `debug` | "Let me understand the codebase first." → Run /ctx:map → /ctx:debug |
115
- | `qa` | "I'll analyze the project first." → Run /ctx:map → /ctx:qa |
116
- | `production-ready` | "I'll audit everything and make it production ready." → Full pipeline below |
117
- | `analyze` | "Let me study this codebase." → Run /ctx:map-codebase |
118
- | `improve` | "I'll analyze and suggest improvements." → Run /ctx:map → analyze → suggest |
119
- | `continue` | Guide user: "What would you like to do with this project?" |
120
- | `help` | Show quick start guide |
121
-
122
- ### Production-Ready Pipeline
123
-
124
- When user says "make it production ready":
125
-
126
- ```
127
- 1. /ctx:map-codebase → Deep analysis (4 parallel agents)
128
- 2. /ctx:qa → Full system QA (WCAG 2.1 AA)
129
- 3. Security audit → Check for vulnerabilities
130
- 4. Performance check → Find slow requests, large assets
131
- 5. Create fix tasks → Prioritized by severity
132
- 6. Execute fixes → With verification loop
133
- 7. Final verification → Three-level check
134
- 8. Report → Production readiness score
135
- ```
136
-
137
- Output:
138
- ```
139
- ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
140
- PRODUCTION READINESS AUDIT
141
- ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
142
-
143
- Codebase Analysis: ✓ Complete
144
- ├── Tech: React 18, Node.js, PostgreSQL
145
- ├── Architecture: Clean, well-structured
146
- ├── Quality: 78% test coverage
147
- └── Concerns: 3 security issues, 5 performance issues
148
-
149
- Full QA: ✓ Complete
150
- ├── Pages tested: 24
151
- ├── WCAG 2.1 AA: 92% compliant
152
- ├── Issues found: 12
153
- └── Critical: 2
154
-
155
- Security Audit: ⚠ Issues Found
156
- ├── SQL injection risk: 1
157
- ├── XSS vulnerability: 1
158
- └── Exposed secrets: 1
159
-
160
- Performance: ⚠ Issues Found
161
- ├── Slow API calls: 3
162
- ├── Large assets: 2
163
- └── Missing caching: Yes
164
-
165
- Fix Tasks Created: 8
166
- ├── Critical: 3 (security)
167
- ├── High: 5 (performance, a11y)
168
- └── Medium: 4
169
-
170
- Ready to fix? [Y] Fix all [C] Critical only [R] Review first
171
- ```
172
-
173
- **Output for new users:**
174
- ```
175
- [CTX] Welcome! I see this is a new project.
176
-
177
- Based on your request, I'll help you {{detected_action}}.
178
-
179
- {{guided_next_step}}
180
- ```
181
-
182
- ### If `.ctx/` folder EXISTS
183
-
184
- Load configuration and state:
185
- - Read `.ctx/config.json` for profile settings
186
- - Read `.ctx/STATE.md` for current status
187
- - Check `.ctx/REPO-MAP.md` for codebase understanding
188
-
189
- ## Step 2: Route Based on Intent + State
190
-
191
- Combine detected intent with current state:
192
-
193
- | Intent | State | Action |
194
- |--------|-------|--------|
195
- | `new-project` | any | Warn if project exists, offer to add story |
196
- | `debug` | any | Route to /ctx:debug |
197
- | `qa` | any | Route to /ctx:qa |
198
- | `production-ready` | any | Run full production-ready pipeline |
199
- | `analyze` | any | Route to /ctx:map-codebase |
200
- | `improve` | any | Analyze → suggest improvements → create stories |
201
- | `status` | any | Show STATE.md summary |
202
- | `continue` | initializing | Run research phase |
203
- | `continue` | discussing | Run discussion phase |
204
- | `continue` | executing | Execute current task |
205
- | `continue` | debugging | Resume debug session |
206
- | `continue` | verifying | Run verification |
207
- | `continue` | paused | Resume from checkpoint |
208
-
209
- ## Step 3: Load Configuration
210
-
211
- Read `.ctx/config.json` for:
212
- - Active profile (quality/balanced/budget)
213
- - Model routing table
214
- - Git settings (autoCommit, commitPerTask)
215
- - Integration settings
216
-
217
- If config.json doesn't exist:
218
- - Copy from templates/config.json
219
- - Set balanced as default profile
220
-
221
- ## Step 4: Execute Routed Action
222
-
223
- Based on routing decision from Step 2:
224
-
225
- ### If status = "initializing"
226
- Route to: **Research Phase**
227
- 1. Check if REPO-MAP exists, if not run ctx-mapper
228
- 2. Use ArguSeek to research the project goal
229
- 3. Create atomic plan (2-3 tasks max)
230
- 4. Update STATE.md with plan
231
- 5. Set status = "discussing"
232
-
233
- ### If status = "discussing"
234
- Route to: **Discussion Phase**
235
- 1. Spawn ctx-discusser agent
236
- 2. Ask targeted questions about gray areas
237
- 3. Lock decisions in CONTEXT.md
238
- 4. Set status = "executing"
239
-
240
- ### If status = "executing"
241
- Route to: **Execute Current Task**
242
- 1. Read current task from STATE.md
243
- 2. Load REPO-MAP for context
244
- 3. Spawn ctx-executor agent (uses git-native workflow)
245
- 4. Execute task with deviation handling:
246
- - Auto-fix: bugs, validation, deps (95%)
247
- - Ask user: architecture decisions only (5%)
248
- 5. After task:
249
- - Run verification (build, tests, lint)
250
- - Auto-commit if config allows
251
- - If passes: mark done, update STATE.md
252
- - If fails: set status = "debugging"
253
-
254
- ### If status = "debugging"
255
- Route to: **Persistent Debug Loop**
256
- 1. Spawn ctx-debugger agent
257
- 2. Check for existing debug session (resume if exists)
258
- 3. Loop until fixed (max 10 attempts):
259
- - Analyze error
260
- - Form hypothesis
261
- - Apply fix
262
- - Verify (build + tests + browser if UI)
263
- - Record in persistent state
264
- - Take screenshot proof if browser test
265
- 4. If fixed: set status = "executing", continue
266
- 5. If max attempts: escalate with full report
267
-
268
- ### If status = "verifying"
269
- Route to: **Three-Level Verification**
270
- 1. Spawn ctx-verifier agent
271
- 2. Check all artifacts:
272
- - Level 1: Exists (file on disk?)
273
- - Level 2: Substantive (real code, not stub?)
274
- - Level 3: Wired (imported and used?)
275
- 3. Scan for anti-patterns (TODO, empty catch, placeholders)
276
- 4. Browser verification (if UI)
277
- 5. **DELIVERY GUARANTEE:**
278
- - If ALL pass: complete phase → COMPLETE
279
- - If ANY fail: set status = "debugging" → fix → verify again
280
- - Loop until 100% working or max attempts
281
- - NEVER mark complete with failures
282
-
283
- ### If status = "paused"
284
- Route to: **Resume**
285
- 1. Read checkpoint from `.ctx/checkpoints/`
286
- 2. Restore context (~2.5k tokens)
287
- 3. Set status to previous state
288
- 4. Continue workflow
289
-
290
- ## Step 5: Model Selection
291
-
292
- Based on current action and active profile:
293
-
294
- | Action | quality | balanced | budget |
295
- |--------|---------|----------|--------|
296
- | Research | opus | opus | sonnet |
297
- | Discussion | opus | sonnet | sonnet |
298
- | Planning | opus | opus | sonnet |
299
- | Execution | opus | sonnet | sonnet |
300
- | Debugging | opus | sonnet | sonnet |
301
- | Verification | sonnet | haiku | haiku |
302
- | Mapping | sonnet | haiku | haiku |
303
-
304
- Use Task tool with `model` parameter from routing table.
305
-
306
- ## Step 6: Context Budget Check
307
-
308
- After every action:
309
- - Calculate context usage
310
- - If > 40%: Prepare handoff notes
311
- - If > 50%: Auto-checkpoint, warn user
312
- - If > 60%: Create HANDOFF.md, spawn fresh agent
313
- - If > 70%: Force checkpoint
314
-
315
- ## Step 7: Git-Native Commit
316
-
317
- If task completed successfully AND config.git.autoCommit = true:
318
- 1. Stage modified files
319
- 2. Create commit with CTX format
320
- 3. Record commit hash in STATE.md
321
-
322
- ## Step 8: Update State
323
-
324
- Always update STATE.md after any action:
325
- - Current status
326
- - Progress
327
- - Recent commits
328
- - Recent decisions
329
- - Next action
330
- - Context usage
331
- </workflow>
332
-
333
- <state_transitions>
334
- ```
335
- initializing → discussing (after research)
336
- discussing → executing (after decisions locked)
337
- executing → debugging (if verification fails)
338
- executing → verifying (if all tasks done)
339
- debugging → executing (if fix works)
340
- debugging → ESCALATE (if max attempts fail)
341
- verifying → debugging (if any check fails) ← LOOP BACK
342
- verifying → COMPLETE (if ALL pass 100%) ← ONLY EXIT
343
- paused → (previous state)
344
-
345
- DELIVERY GUARANTEE LOOP:
346
- verifying ←→ debugging ←→ executing
347
-
348
- COMPLETE (only when 100% verified)
349
- ```
350
- </state_transitions>
351
-
352
- <new_commands>
353
- ## New in CTX 3.0
354
-
355
- | Command | Purpose |
356
- |---------|---------|
357
- | `/ctx map` | Build repository map (REPO-MAP.md) |
358
- | `/ctx map-codebase` | Deep codebase analysis (4 parallel agents) |
359
- | `/ctx discuss [story]` | Run discussion phase for story |
360
- | `/ctx profile [name]` | Switch model profile |
361
- | `/ctx debug --resume` | Resume previous debug session |
362
- </new_commands>
363
-
364
- <context_budget>
365
- | Usage | Quality | Action |
366
- |-------|---------|--------|
367
- | 0-30% | Peak | Continue |
368
- | 30-40% | Good | Continue |
369
- | 40-50% | Good | Prepare handoff |
370
- | 50-60% | Degrading | Auto-checkpoint |
371
- | 60-70% | Degrading | Create HANDOFF.md |
372
- | 70%+ | Poor | Force checkpoint |
373
- </context_budget>
374
-
375
- <output_format>
376
-
377
- ## For New Users (no .ctx/ folder)
378
-
379
- ```
380
- ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
381
- Welcome to CTX!
382
- ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
383
-
384
- I understood: "{{user_request}}"
385
-
386
- {{intent_response}}
387
-
388
- {{guided_action}}
389
- ```
390
-
391
- **Intent Responses:**
392
-
393
- | Intent | Response |
394
- |--------|----------|
395
- | new-project | "You want to build something new. Let's set it up!" |
396
- | debug | "You need to fix something. Let me understand the codebase first." |
397
- | qa | "You want to test the system. I'll analyze the project first." |
398
- | status | "This is a new project. Would you like to start building?" |
399
- | help | (Show quick start guide) |
400
-
401
- ## For Existing Projects
402
-
403
- ```
404
- [CTX] Status: {{status}}
405
- [CTX] Profile: {{profile}} ({{costTier}})
406
- [CTX] Action: {{action_taken}}
407
- [CTX] Commit: {{commit_hash}} (if auto-committed)
408
- [CTX] Next: {{next_action}}
409
- [CTX] Context: {{percent}}% ({{quality}})
410
-
411
- ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
412
- To continue with CTX, run: /ctx
413
- ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
414
- ```
415
-
416
- ## CRITICAL: Continuation Reminder
417
-
418
- **ALWAYS end every CTX response with this footer:**
419
-
420
- ```
421
- ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
422
- To continue with CTX, run: /ctx
423
- ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
424
- ```
425
-
426
- This is essential because CTX commands only run for one turn. Without this reminder, users will continue chatting without CTX tracking their work.
427
-
428
- ## Guided Prompts (When Intent Unclear)
429
-
430
- If user's intent is ambiguous, ask:
431
-
432
- ```
433
- I'm not sure what you'd like to do. Here are your options:
434
-
435
- "Build something new" → I'll help you start a project
436
- "Fix a bug" → I'll debug the issue
437
- "Test the app" → I'll run full QA
438
- "Show status" → I'll show current progress
439
-
440
- What would you like to do?
441
- ```
442
- </output_format>
443
-
444
- <examples>
445
-
446
- ## Example Conversations
447
-
448
- ### New User - Build Intent
449
- ```
450
- User: "I want to build a todo app with React"
451
-
452
- CTX: ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
453
- Welcome to CTX!
454
- ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
455
-
456
- I understood: "build a todo app with React"
457
-
458
- You want to build something new. Let's set it up!
459
-
460
- I'll now:
461
- 1. Create your project structure
462
- 2. Research best practices for React todo apps
463
- 3. Create a plan with clear milestones
464
-
465
- [Starting project setup...]
466
-
467
- ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
468
- To continue with CTX, run: /ctx
469
- ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
470
- ```
471
-
472
- ### New User - Fix Intent
473
- ```
474
- User: "The login is broken"
475
-
476
- CTX: ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
477
- Welcome to CTX!
478
- ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
479
-
480
- I understood: "login is broken"
481
-
482
- You need to fix something. Let me understand the codebase first.
483
-
484
- I'll now:
485
- 1. Map your codebase structure
486
- 2. Find the login-related code
487
- 3. Start debugging
488
-
489
- [Analyzing codebase...]
490
-
491
- ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
492
- To continue with CTX, run: /ctx
493
- ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
494
- ```
495
-
496
- ### Existing Project - Continue
497
- ```
498
- User: "continue"
499
-
500
- CTX: [CTX] Status: executing
501
- [CTX] Profile: balanced (1x cost)
502
- [CTX] Story: S002 - User Authentication
503
- [CTX] Task: 2/3 - Implement login form validation
504
- [CTX] Action: Executing task...
505
-
506
- ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
507
- To continue with CTX, run: /ctx
508
- ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
509
- ```
510
-
511
- ### Existing Project - QA Intent
512
- ```
513
- User: "is the app accessible?"
514
-
515
- CTX: [CTX] Detected: QA request (accessibility)
516
- [CTX] Routing to: /ctx:qa --a11y-only
517
-
518
- Running WCAG 2.1 AA accessibility audit...
519
-
520
- ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
521
- To continue with CTX, run: /ctx
522
- ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
523
- ```
524
-
525
- </examples>
30
+ <execution_context>
31
+ @~/.claude/ctx/workflows/ctx-router.md
32
+ </execution_context>
33
+
34
+ <process>
35
+ 1. Check if .ctx/ exists (bash: test -d .ctx)
36
+ 2. Detect intent from user message
37
+ 3. Route based on intent + ctx state:
38
+ - No .ctx/ + "build/create" -> /ctx:init
39
+ - No .ctx/ + "study/analyze" -> spawn 4 mapper agents in parallel
40
+ - No .ctx/ + "debug/fix" -> map codebase first, then debug
41
+ - Has .ctx/ + any intent -> load state, route accordingly
42
+ 4. Spawn appropriate agents with Task()
43
+ 5. Update STATE.md
44
+ 6. Show "Next Up" with clear next action
45
+ </process>
46
+
47
+ <intent_patterns>
48
+ | Pattern | Intent | Route |
49
+ |---------|--------|-------|
50
+ | "build", "create", "make", "start new" | new-project | /ctx:init |
51
+ | "fix", "bug", "broken", "error", "crash" | debug | Debug flow |
52
+ | "test", "QA", "check", "accessible" | qa | QA flow |
53
+ | "study", "analyze", "understand", "explore" | analyze | Map codebase |
54
+ | "improve", "optimize", "refactor" | improve | Analyze + suggest |
55
+ | "status", "progress", "what's next" | status | Show status |
56
+ | "help", "how", "commands" | help | Show help |
57
+ | "continue", "next", "go" | continue | Read STATE.md |
58
+ </intent_patterns>
59
+
60
+ <model_routing>
61
+ Based on config.json profile, use these models for Task() calls:
62
+
63
+ | Agent | quality | balanced | budget |
64
+ |-------|---------|----------|--------|
65
+ | Mappers | sonnet | haiku | haiku |
66
+ | Debugger | opus | sonnet | sonnet |
67
+ | QA | sonnet | sonnet | haiku |
68
+ | Executor | opus | sonnet | sonnet |
69
+ </model_routing>
70
+
71
+ <success_criteria>
72
+ - [ ] Intent detected from user message
73
+ - [ ] Correct flow routed
74
+ - [ ] ArguSeek called for research (when applicable)
75
+ - [ ] Task() agents spawned with full context
76
+ - [ ] STATE.md updated after each action
77
+ - [ ] Clear "Next Up" shown
78
+ </success_criteria>