opencode-orchestrator 0.8.15 → 0.8.17

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.
Files changed (2) hide show
  1. package/README.md +84 -424
  2. package/package.json +1 -1
package/README.md CHANGED
@@ -12,449 +12,133 @@
12
12
 
13
13
  ---
14
14
 
15
- ## Overview
15
+ ## ⚡ Quick Start
16
16
 
17
- Multi-agent system that autonomously executes complex tasks. Commander delegates work to specialized agents, manages parallel execution, and maintains shared context.
18
-
19
- | Capability | Detail |
20
- |------------|--------|
21
- | **Agents** | 4 consolidated (Commander, Planner, Worker, Reviewer) |
22
- | **Parallel Sessions** | Up to 50 concurrent Worker Sessions |
23
- | **Context Management** | `.opencode/` with adaptive summarization |
24
- | **Memory Safety** | Auto GC, disk archiving |
25
- | **Error Handling** | Pattern-based auto recovery + session recovery |
26
- | **Todo Continuation** | Auto-continues when todos remain |
27
- | **Smart Notifications** | TaskToastManager with consolidated views |
28
-
29
- ---
30
-
31
- ## 🏛️ Architecture Overview
32
-
33
- ### 🚀 User Flow: Real Scenario
34
-
35
- > **Example:** `/task "Build a REST API with user authentication"`
36
-
37
- ```
38
- ┌─────────────────────────────────────────────────────────────────────────────────┐
39
- │ 👤 USER: /task "Build a REST API with user authentication" │
40
- └─────────────────────────────────────────────────────────────────────────────────┘
41
- │
42
- ▼
43
- ┌─────────────────────────────────────────────────────────────────────────────────┐
44
- │ 🎯 COMMANDER (Master Session) │
45
- │ │
46
- │ "I'll break this into parallel tasks and delegate to specialized agents" │
47
- │ │
48
- │ ┌─────────────────────────────────────────────────────────────────────────┐ │
49
- │ │ 📋 Creates TODO.md: │ │
50
- │ │ - [ ] Research: Express.js + JWT best practices │ │
51
- │ │ - [ ] Setup: Project structure + dependencies │ │
52
- │ │ - [ ] Implement: User model + auth routes │ │
53
- │ │ - [ ] Test: API endpoints verification │ │
54
- │ └─────────────────────────────────────────────────────────────────────────┘ │
55
- └─────────────────────────────────────────────────────────────────────────────────┘
56
- │
57
- ┌────────────────────────┼────────────────────────┐
58
- │ │ │
59
- ▼ ▼ ▼
60
-
61
- ╔═══════════════════╗ ╔═══════════════════╗ ╔═══════════════════╗
62
- ║ 🔍 PLANNER ║ ║ 🔨 WORKER #1 ║ ║ 🔨 WORKER #2 ║
63
- ║ Session: ses_001 ║ ║ Session: ses_002 ║ ║ Session: ses_003 ║
64
- ╠═══════════════════╣ ╠═══════════════════╣ ╠═══════════════════╣
65
- ║ Research Express ║ ║ Create src/ ║ ║ Write auth.ts ║
66
- ║ + JWT docs ║ ║ Setup package.json║ ║ + user.model.ts ║
67
- ║ ║ ║ ║ ║ ║
68
- ║ 📥 websearch() ║ ║ 📥 write files ║ ║ 📥 write files ║
69
- ║ ⏳ cache_docs() ║ ║ ⏳ run_background ║ ║ ║
70
- ╚═══════════════════╝ ║ → npm install ║ ╚═══════════════════╝
71
- │ ╚═══════════════════╝ │
72
- │ │ │
73
- │ ⚡ ALL RUN IN PARALLEL (async: true) │
74
- │ ⏱️ Commander monitors, doesn't wait │
75
- └────────────────────────┼──────────────────────┘
76
- ▼
77
- ┌─────────────────────────────────────────────────────────────────────────────────┐
78
- │ 🔄 BACKGROUND PROCESSES │
79
- │ │
80
- │ ┌─────────────────┐ ┌─────────────────┐ ┌─────────────────┐ │
81
- │ │ npm install │ │ npm run build │ │ npm test │ │
82
- │ │ (bg_cmd_001) │ │ (bg_cmd_002) │ │ (bg_cmd_003) │ │
83
- │ │ ⏳ running... │ │ ⏳ running... │ │ ⏳ pending... │ │
84
- │ └─────────────────┘ └─────────────────┘ └─────────────────┘ │
85
- │ │
86
- │ • Check with: check_background / list_background │
87
- │ • Non-blocking: Commander continues other work │
88
- └─────────────────────────────────────────────────────────────────────────────────┘
89
- │
90
- ▼
91
- ┌─────────────────────────────────────────────────────────────────────────────────┐
92
- │ 📊 ASYNC RESULT COLLECTION │
93
- │ │
94
- │ get_task_result("ses_001") ──▶ ✅ Planner done: "Found JWT patterns..." │
95
- │ get_task_result("ses_002") ──▶ ✅ Worker #1 done: "Project setup complete" │
96
- │ get_task_result("ses_003") ──▶ ⏳ Still running... │
97
- │ │
98
- │ Commander: "Worker #2 still working, I'll delegate review task meanwhile" │
99
- └─────────────────────────────────────────────────────────────────────────────────┘
100
- │
101
- ▼
102
- ╔═══════════════════╗
103
- ║ ✅ REVIEWER ║
104
- ║ Session: ses_004 ║
105
- ╠═══════════════════╣
106
- ║ Verify endpoints ║
107
- ║ Update TODO.md: ║
108
- ║ - [x] Research ✓ ║
109
- ║ - [x] Setup ✓ ║
110
- ║ - [x] Implement ✓║
111
- ║ - [x] Test ✓ ║
112
- ╚═══════════════════╝
113
- │
114
- ▼
115
- ┌─────────────────────────────────────────────────────────────────────────────────┐
116
- │ 🎖️ MISSION COMPLETE │
117
- │ │
118
- │ Commander: "All TODOs checked. Mission complete." │
119
- │ │
120
- │ Output: <mission_seal>SEALED</mission_seal> │
121
- │ │
122
- │ ✅ Session cleanup → Resources freed → Ready for next mission │
123
- └─────────────────────────────────────────────────────────────────────────────────┘
124
- ```
125
-
126
- **Key Concepts Visualized:**
127
-
128
- | Concept | How It Works |
129
- |---------|--------------|
130
- | **Session** | Each agent runs in isolated session (ses_001, ses_002...) |
131
- | **Parallel Agents** | Multiple Workers execute simultaneously, not sequentially |
132
- | **Background Commands** | Long-running commands (npm install) don't block agents |
133
- | **Async Collection** | Commander polls results, continues work while waiting |
134
- | **Auto-Continue** | If TODOs remain, loop continues until all `[x]` checked |
135
-
136
- ### 📊 Project Architecture
137
-
138
- ```
139
- ┌─────────────────────────────────────────────────────────────────────────────────┐
140
- │ OpenCode Orchestrator Plugin │
141
- │ src/index.ts │
142
- └─────────────────────────────────────────────────────────────────────────────────┘
143
- │
144
- ┌───────────────────────────┼───────────────────────────┐
145
- ▼ ▼ ▼
146
- ┌─────────────────────┐ ┌─────────────────────┐ ┌─────────────────────────────┐
147
- │ Plugin Hooks │ │ Core Systems │ │ Tools │
148
- │ plugin-handlers/ │ │ src/core/ │ │ src/tools/ │
149
- ├─────────────────────┤ ├─────────────────────┤ ├─────────────────────────────┤
150
- │ • chat.message │ │ • agents/ │ │ • parallel/ │
151
- │ • event │ │ ParallelManager │ │ delegate_task │
152
- │ • tool.execute │ │ TaskStore │ │ get_task_result │
153
- │ • assistant.done │ │ Concurrency │ │ • web/ │
154
- └─────────────────────┘ │ • loop/ │ │ webfetch, websearch │
155
- │ MissionSeal │ │ • background-cmd/ │
156
- │ TodoContinuation │ │ run_background │
157
- │ • recovery/ │ │ • search │
158
- │ SessionRecovery │ │ grep, glob, mgrep │
159
- │ • notification/ │ └─────────────────────────────┘
160
- │ Toast, Manager │
161
- │ • session/ │
162
- │ SharedContext │
163
- │ • cache/ │
164
- │ DocumentCache │
165
- └─────────────────────┘
166
- │
167
- ┌───────────────────────────┼───────────────────────────┐
168
- ▼ ▼ ▼
169
- ┌─────────────────────┐ ┌─────────────────────┐ ┌─────────────────────┐
170
- │ 🎯 Commander │ │ 📋 Planner │ │ 🔨 Worker │
171
- │ Orchestrator │ │ Research+Plan │ │ Implementation │
172
- ├─────────────────────┤ ├─────────────────────┤ ├─────────────────────┤
173
- │ • Mission control │ │ • Task decompose │ │ • Code writing │
174
- │ • Parallel delegate │ │ • Doc research │ │ • File operations │
175
- │ • TODO monitoring │ │ • TODO creation │ │ • Command execution │
176
- └─────────────────────┘ └─────────────────────┘ └─────────────────────┘
177
- │ │ │
178
- │ │ │
179
- │ ┌────────┴────────┐ │
180
- │ ▼ │ │
181
- │ ┌─────────────────────┐ │ │
182
- │ │ ✅ Reviewer │ │ │
183
- │ │ Quality+Context │ │ │
184
- │ ├─────────────────────┤ │ │
185
- │ │ • Verification │ │ │
186
- │ │ • TODO updates │ │ │
187
- │ │ • Context manage │ │ │
188
- │ └─────────────────────┘ │ │
189
- │ │ │ │
190
- └──────────────────┼─────────────────┴──────────────────┘
191
- ▼
192
- ┌─────────────────────────────────────────────────────────────────────────────────┐
193
- │ 📁 Shared Workspace (.opencode/) │
194
- ├─────────────────────────────────────────────────────────────────────────────────┤
195
- │ todo.md │ Hierarchical task list (Epic → Task → Subtask) │
196
- │ context.md │ Adaptive context (shrinks with progress) │
197
- │ loop-state.json │ Mission loop iteration state │
198
- │ docs/ │ Cached documentation (auto-expire) │
199
- │ archive/ │ Old context snapshots │
200
- └─────────────────────────────────────────────────────────────────────────────────┘
201
- ```
202
-
203
-
204
- ### 📂 Directory Structure
205
-
206
- ```
207
- src/
208
- ├── index.ts # Plugin main entry point
209
- ├── agents/
210
- │ ├── commander.ts # Commander agent definition
211
- │ ├── definitions.ts # Agent exports
212
- │ ├── prompts/ # Prompt fragments (commander/, planner/, worker/, reviewer/)
213
- │ └── subagents/ # Subagent definitions (planner.ts, worker.ts, reviewer.ts)
214
- ├── core/
215
- │ ├── agents/ # ParallelAgentManager, TaskStore, ConcurrencyController
216
- │ ├── cache/ # DocumentCache
217
- │ ├── loop/ # MissionSeal, TodoContinuation, TodoEnforcer
218
- │ ├── notification/ # Toast, TaskToastManager
219
- │ ├── recovery/ # SessionRecovery, ErrorPatterns
220
- │ ├── session/ # SharedContext
221
- │ └── task/ # TaskScheduler, TaskParser
222
- ├── plugin-handlers/
223
- │ ├── chat-message-handler.ts # /task detection, mission start
224
- │ ├── event-handler.ts # session.idle, session.error handling
225
- │ ├── tool-execute-handler.ts # Tool completion tracking
226
- │ └── assistant-done-handler.ts # Response completion
227
- ├── shared/
228
- │ ├── constants/ # PATHS, TOOL_NAMES, MISSION_SEAL, etc.
229
- │ ├── agent/ # Agent definitions, names
230
- │ └── errors/ # Error types
231
- ├── tools/
232
- │ ├── parallel/ # delegate_task, get_task_result, list_tasks, cancel_task
233
- │ ├── web/ # webfetch, websearch, codesearch, cache_docs
234
- │ ├── background-cmd/ # run_background, check_background, list_background
235
- │ └── search.ts # grep_search, glob_search, mgrep
236
- └── utils/ # Utility functions
17
+ ```bash
18
+ npm install -g opencode-orchestrator
237
19
  ```
238
20
 
239
-
240
- ## Installation
21
+ Restart OpenCode after installation. Then:
241
22
 
242
23
  ```bash
243
- npm install -g opencode-orchestrator
24
+ /task "Build a REST API with authentication"
244
25
  ```
245
26
 
246
- Restart OpenCode after installation.
27
+ That's it! The Commander takes over and works until the mission is complete.
247
28
 
248
29
  ---
249
30
 
250
- ## Usage
31
+ ## 🎯 What is This?
251
32
 
252
- ### 🚀 Two Modes of Operation
253
-
254
- | Mode | Trigger | Behavior |
255
- |------|---------|----------|
256
- | **Commander Mode** 🎯 | `/task "mission"` | Full autonomous execution until sealed |
257
- | **Chat Mode** 💬 | Regular conversation | Simple Q&A, no autonomous behavior |
33
+ A multi-agent system that **autonomously executes complex tasks** from start to finish. Just describe your mission — the Commander handles everything else.
258
34
 
259
35
  ---
260
36
 
261
- ### 🎯 Commander Mode - `/task` (Recommended for Real Work)
262
-
263
- Use `/task` when you need the AI to **complete a mission autonomously**:
37
+ ## ✨ Key Features
264
38
 
265
- ```bash
266
- /task "Fix the login bug in the docker-compose environment"
267
- /task "Add dark mode support to the entire app"
268
- /task "Refactor the API to use TypeScript"
269
- ```
270
-
271
- **What Commander Mode Does:**
272
- - ♾️ **Runs until sealed** — Loops until agent outputs `<mission_seal>SEALED</mission_seal>`
273
- - 🧠 **Anti-Hallucination** — Researches docs before coding
274
- - ⚡ **Parallel Execution** — Up to 50 concurrent Worker Sessions
275
- - 🔄 **Auto-Recovery** — Handles errors automatically with pattern matching
276
- - 📊 **Progress Tracking** — Monitors TODO completion and shows progress
277
-
278
- **Concurrency Limits (per agent type):**
279
- | Agent | Max Concurrent | Purpose |
280
- |--------|---------------|-----------|
281
- | Commander | 1 | Single orchestrator per mission |
282
- | Planner | 3 | Research and TODO planning |
283
- | Worker | 10 | Implementation tasks |
284
- | Reviewer | 5 | Verification and testing |
285
-
286
- **🎖️ Mission Seal Loop:**
287
- ```
288
- /task "mission" → Agent works → Idle? → Seal found?
289
- ↑ │
290
- │ No │ Yes
291
- └──────────────┴──→ ✅ Complete
292
- ```
293
-
294
- When the agent finishes ALL work, it outputs:
295
- ```xml
296
- <mission_seal>SEALED</mission_seal>
297
- ```
298
-
299
- **Control Commands:**
300
- - `/stop` or `/cancel` — Stop the loop manually
301
- - Max 20 iterations (configurable)
302
-
303
- <div align="center">
304
- <img src="assets/tui_image.png" alt="Commander TUI" width="600" />
305
- <p><sub><b>/task "mission"</b> triggers full Commander mode with Mission Seal loop</sub></p>
306
- </div>
39
+ | Feature | Description |
40
+ |---------|-------------|
41
+ | 🤖 **4 Specialized Agents** | Commander → Planner → Worker → Reviewer, each with distinct roles |
42
+ | ⚡ **Parallel Execution** | Up to 50 concurrent sessions — no waiting, maximum throughput |
43
+ | 🔄 **Background Tasks** | Long-running commands (npm install, tests) run non-blocking |
44
+ | 📋 **TODO-Driven Completion** | Auto-continues until every TODO is checked — never stops halfway |
45
+ | 🛡️ **Auto-Recovery** | Handles errors, rate limits, and session crashes automatically |
46
+ | 💾 **Memory Safety** | Auto GC, disk archiving, 60-min TTL — no memory leaks |
47
+ | 🔔 **Smart Notifications** | Consolidated task toasts with progress tracking |
307
48
 
308
49
  ---
309
50
 
310
- ### 💬 Chat Mode - Regular Conversation (Simple Q&A)
311
-
312
- Just type normally without `/task` for simple questions:
51
+ ## 🚀 How It Works
313
52
 
314
53
  ```
315
- How do I add a loading spinner?
316
- What's the difference between useState and useReducer?
54
+ 👤 You: /task "Build a REST API with authentication"
55
+ │
56
+ ▼
57
+ ┌───────────────────────────────┐
58
+ │ 🎯 COMMANDER │ ← Master Session
59
+ │ Creates TODO, delegates │
60
+ └───────────────────────────────┘
61
+ │
62
+ ┌──────────────────────┼──────────────────────┐
63
+ ▼ ▼ ▼
64
+ ┌─────────┐ ┌─────────┐ ┌─────────┐
65
+ │ PLANNER │ │ WORKER │ │ WORKER │
66
+ │ ses_001 │ │ ses_002 │ │ ses_003 │ ⚡ PARALLEL
67
+ └─────────┘ └─────────┘ └─────────┘ SESSIONS
68
+ │ │ │
69
+ │ ┌─────────────────┴─────────────────┐ │
70
+ │ │ 🔄 Background: npm install, test │ │ ⏳ NON-BLOCKING
71
+ │ └───────────────────────────────────┘ │
72
+ └──────────────────────┼──────────────────────┘
73
+ ▼
74
+ ┌───────────────────────────────┐
75
+ │ ✅ REVIEWER │
76
+ │ Verifies, updates TODOs │
77
+ └───────────────────────────────┘
78
+ │
79
+ ┌─────────┴─────────┐
80
+ │ TODOs remain? │
81
+ └─────────┬─────────┘
82
+ Yes ↓ ↓ No
83
+ ♻️ LOOP 🎖️ SEALED
84
+ BACK (done!)
317
85
  ```
318
86
 
319
- **Chat Mode is just regular conversation** — no autonomous execution, no parallel agents, no mission tracking.
87
+ **♾️ Runs until complete** — Keeps looping until ALL TODOs are checked, then seals the mission.
320
88
 
321
89
  ---
322
90
 
323
- > **💡 Pro Tip:** Use `/task` for anything that requires multiple steps, file changes, or verification. Use Chat Mode for quick questions.
324
-
325
- ---
91
+ ## 🎮 Usage
326
92
 
327
- ## 🔄 Mission Loop Mechanism
93
+ | Mode | Trigger | What Happens |
94
+ |------|---------|--------------|
95
+ | **Commander Mode** 🎯 | `/task "mission"` | Full autonomous execution |
96
+ | **Chat Mode** 💬 | Just type normally | Simple Q&A, no automation |
328
97
 
329
- The Commander uses an event-driven mission loop to autonomously complete missions:
330
-
331
- ```
332
- ┌─────────────────────────────────────────────────────────────────┐
333
- │ MISSION LOOP LIFECYCLE │
334
- └─────────────────────────────────────────────────────────────────┘
335
-
336
- 1️⃣ User sends "/task 'mission'"
337
- ↓
338
- 2️⃣ chat.message handler detects /task
339
- ↓
340
- 3️⃣ Create session + startMissionLoop()
341
- ↓
342
- Write .opencode/loop-state.json:
343
- {
344
- "active": true,
345
- "sessionID": "ses_abc",
346
- "iteration": 1,
347
- "maxIterations": 20
348
- }
349
- ↓
350
- 4️⃣ Commander receives prompt → delegates work
351
- ↓
352
- 5️⃣ Worker sessions execute → results collected
353
- ↓
354
- 6️⃣ session.idle event triggers
355
- ↓
356
- Check for <mission_seal>SEALED</mission_seal>
357
- ├─ Seal found? → Clear loop state → Complete ✅
358
- └─ Not found? → Increment iteration → Continue loop
359
- ↓
360
- Show countdown toast (3 seconds)
361
- ↓
362
- Inject continuation prompt
363
- ↓
364
- [Loop back to step 4]
98
+ **Example Commands:**
99
+ ```bash
100
+ /task "Fix the login bug in docker-compose"
101
+ /task "Add dark mode to the entire app"
102
+ /task "Refactor the API to TypeScript"
365
103
  ```
366
104
 
367
- **Key Loop Components:**
368
-
369
- | Component | File | Purpose |
370
- |-----------|-------|---------|
371
- | Loop State | `src/core/loop/mission-seal.ts` | State management (.opencode/loop-state.json) |
372
- | Seal Detection | `src/core/loop/mission-seal-handler.ts` | Detect `<mission_seal>` in responses |
373
- | Continuation | `src/core/loop/mission-seal-handler.ts` | Inject prompts to continue work |
374
- | Countdown | `src/core/loop/mission-seal-handler.ts` | 3-second countdown toast |
375
- | Idle Handler | `src/plugin-handlers/assistant-done-handler.ts` | Monitor session.idle events |
105
+ **Stop anytime:** `/stop` or `/cancel`
376
106
 
377
- **Why Event-Driven?**
378
- - No fixed iteration limits - loop continues until sealed
379
- - Resilient to network delays
380
- - Can be interrupted by user at any time
381
- - Efficient polling (500ms interval with backoff)
382
-
383
- ---
384
-
385
- ## The 4 Agents
386
-
387
- | Agent | Role | Responsibility |
388
- | :--------------- | :----------- | :--------------------------------- |
389
- | **Commander** 🎯 | Orchestrator | Autonomous mission control |
390
- | **Planner** 📋 | Planner + Research | Task decomposition, research, caching docs |
391
- | **Worker** 🔨 | Developer + Docs | Full-stack implementation, documentation |
392
- | **Reviewer** ✅ | Quality + Context | Verification, TODO updates, context management |
393
-
394
- ---
395
-
396
- ## 🛠️ Available Tools
397
-
398
- | Tool | Description |
399
- |------|-------------|
400
- | `delegate_task` | Delegate work to subagent (async/sync) |
401
- | `get_task_result` | Get result from delegated task |
402
- | `list_tasks` | List all running/completed tasks |
403
- | `cancel_task` | Cancel a running task |
404
- | `webfetch` | Fetch URL content as Markdown |
405
- | `websearch` | Search web (SearXNG → Brave → DuckDuckGo) |
406
- | `codesearch` | Search open source code patterns |
407
- | `cache_docs` | Manage cached documentation |
408
- | `run_background` | Run command in background |
409
- | `check_background` / `list_background` | Monitor background jobs |
410
- | `grep_search` / `glob_search` / `mgrep` | Fast file search |
107
+ <div align="center">
108
+ <img src="assets/tui_image.png" alt="Commander TUI" width="600" />
109
+ </div>
411
110
 
412
111
  ---
413
112
 
414
- ## 📊 Resource Guarantees
113
+ ## 🤖 The 4 Agents
415
114
 
416
- | Resource | Limit | Safety Mechanism |
417
- |----------|-------|------------------|
418
- | Parallel Sessions | 50 | Queue overflow protection |
419
- | Tasks in Memory | 1,000 | Auto GC + disk archive |
420
- | Notifications | 100/parent | FIFO eviction |
421
- | Event History | 100 | Ring buffer |
422
- | Session TTL | 60 min | Auto cleanup |
423
- | Poll Interval | 500ms | Fast completion detection |
424
- | Max Poll Count | 600 | Hard limit prevents infinite loops |
425
- | Sync Timeout | 5 min | Safe delegate_task timeout |
426
- | Recovery Attempts | 3 | Auto session error recovery |
115
+ | Agent | Role | Key Actions |
116
+ |-------|------|-------------|
117
+ | **Commander** 🎯 | Orchestrator | Delegates tasks, monitors progress, seals mission |
118
+ | **Planner** 📋 | Research + Plan | Web search, doc caching, TODO creation |
119
+ | **Worker** 🔨 | Implementation | Code writing, file ops, background commands |
120
+ | **Reviewer** ✅ | Quality Control | Verification, TODO updates, context management |
427
121
 
428
122
  ---
429
123
 
430
- ## 🧪 Test Coverage
124
+ ## 📊 Resource Limits
431
125
 
432
- ```
433
- Test Files: 19 passed
434
- Tests: 216 passed
435
- Duration: ~4.3s
436
- ```
126
+ | Resource | Limit |
127
+ |----------|-------|
128
+ | Parallel Sessions | 50 |
129
+ | Tasks in Memory | 1,000 (auto GC) |
130
+ | Session TTL | 60 min |
131
+ | Recovery Attempts | 3 per session |
132
+ | Max Iterations | 20 per mission |
437
133
 
438
134
  ---
439
135
 
440
- ## 🔄 Reliability Features
136
+ ## 📚 Documentation
441
137
 
442
- ### Session Recovery
443
- Automatic recovery from common errors:
444
- - `tool_result_missing` - Tool crash recovery
445
- - `thinking_block_order` - Thinking block issues
446
- - `rate_limit` - API rate limiting with backoff
447
- - Max 3 recovery attempts per session
138
+ For detailed architecture, directory structure, and internals:
448
139
 
449
- ### Todo Continuation
450
- - Monitors `session.idle` events
451
- - 3-second countdown toast before auto-continuation
452
- - Cancels on user interaction
453
- - Skips if background tasks running or session is recovering
454
-
455
- ### noReply Optimization
456
- - Individual task completion: `noReply: true` (saves tokens)
457
- - All tasks complete: `noReply: false` (AI processes results)
140
+ - **[System Architecture](docs/SYSTEM_ARCHITECTURE.md)** — Complete technical documentation
141
+ - [Release Notes](docs/releases/) — Version history
458
142
 
459
143
  ---
460
144
 
@@ -466,30 +150,6 @@ npm uninstall -g opencode-orchestrator
466
150
 
467
151
  ---
468
152
 
469
- ## 🔧 Debugging
470
-
471
- **Log file location:**
472
- ```bash
473
- # Find log path (macOS uses /var/folders/...)
474
- npm run log
475
- # or:
476
- tail -f "$(node -e 'console.log(require("os").tmpdir())')/opencode-orchestrator.log"
477
-
478
- # Windows
479
- # C:\Users\<username>\AppData\Local\Temp\opencode-orchestrator.log
480
- ```
481
-
482
- ---
483
-
484
- ## Documentation
485
-
486
- - **[System Architecture](docs/SYSTEM_ARCHITECTURE.md)** — Detailed technical docs
487
- - **[Architecture and Flow](docs/ARCHITECTURE_AND_FLOW.md)** — Complete architecture guide with scenarios
488
- - **[Improvement Suggestions](docs/IMPROVEMENT_SUGGESTIONS.md)** — Project improvement recommendations
489
- - [Release Notes](docs/releases/) — Version history
490
-
491
- ---
492
-
493
153
  ## License
494
154
 
495
155
  MIT License. [LICENSE](LICENSE)
package/package.json CHANGED
@@ -2,7 +2,7 @@
2
2
  "name": "opencode-orchestrator",
3
3
  "displayName": "OpenCode Orchestrator",
4
4
  "description": "Distributed Cognitive Architecture for OpenCode. Turns simple prompts into specialized multi-agent workflows (Planner, Coder, Reviewer).",
5
- "version": "0.8.15",
5
+ "version": "0.8.17",
6
6
  "author": "agnusdei1207",
7
7
  "license": "MIT",
8
8
  "repository": {