opencode-orchestrator 0.9.24 → 0.9.25

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 +60 -117
  2. package/package.json +1 -1
package/README.md CHANGED
@@ -12,75 +12,6 @@
12
12
 
13
13
  ---
14
14
 
15
- ## 🔥 Why This Plugin?
16
-
17
- **Budget models have limits. Orchestration breaks them.**
18
-
19
- ---
20
-
21
- ## 🧠 Core Philosophy
22
-
23
- > **Explore → Adapt → Act**
24
- > *Never assume. Always verify. Then execute.*
25
-
26
- ### The Problem with AI Agents
27
-
28
- Most AI agents fail because they:
29
- - **Assume** instead of checking
30
- - **Guess** instead of researching
31
- - Follow **rigid rules** instead of adapting
32
- - Trust **memory** instead of verifying
33
-
34
- ### Our Solution: Adaptive Intelligence
35
-
36
- Every agent in this system follows a simple principle:
37
-
38
- ```
39
- ┌─────────────────────────────────────────────────────────────┐
40
- │ │
41
- │ 🔍 EXPLORE → 🔄 ADAPT → ⚡ ACT │
42
- │ │
43
- │ Discover Adjust to Execute with │
44
- │ the reality what you find confidence │
45
- │ │
46
- └─────────────────────────────────────────────────────────────┘
47
- ```
48
-
49
- ### Agent Philosophies
50
-
51
- | Agent | Philosophy | What It Means |
52
- |:-----:|:-----------|:--------------|
53
- | 🎯 **Commander** | Explore → Adapt → Act | Discover project structure first, then delegate intelligently |
54
- | 📋 **Planner** | Research → Plan → Document | Never guess syntax. Find official docs, verify, then plan |
55
- | 🔨 **Worker** | Observe → Learn → Implement | Study existing code patterns, then write code that fits |
56
- | ✅ **Reviewer** | Understand → Verify → Approve | Know the project's standards, run actual tests, approve with evidence |
57
-
58
- ### Guiding Principles
59
-
60
- 1. **🔍 Never Assume, Always Discover**
61
- - Read `context.md` before every action
62
- - Detect the tech stack, don't assume it
63
- - Find the build/test commands, don't guess them
64
-
65
- 2. **📚 Evidence Over Memory**
66
- - Cite official documentation for every claim
67
- - Run actual commands for verification
68
- - Mark complete only with proof
69
-
70
- 3. **🔄 Adapt to the Project**
71
- - Match existing code patterns
72
- - Follow discovered conventions
73
- - Respect the project's way of doing things
74
-
75
- 4. **🤝 Separation of Concerns**
76
- - Commander orchestrates, never implements
77
- - Worker builds, never verifies
78
- - Reviewer approves, never codes
79
- - Each agent does one thing excellently
80
-
81
- ---
82
-
83
-
84
15
  ## ⚡ Quick Start
85
16
 
86
17
  ```bash
@@ -88,36 +19,53 @@ npm install -g opencode-orchestrator
88
19
  ```
89
20
 
90
21
  Then in OpenCode:
91
-
92
22
  ```bash
93
23
  /task "Build a REST API with authentication"
94
24
  ```
95
25
 
96
- **Done.** You just deployed an army of AI agents. They plan, build, test, fix — and don't stop until it's sealed. ✨
26
+ **Done.** AI agents plan, build, test, fix — and don't stop until it's sealed. ✨
27
+
28
+ > **Stop anytime:** `/stop`
97
29
 
98
30
  ---
99
31
 
100
- ## 🎯 What Makes This Different?
32
+ ## 🧠 Core Philosophy
33
+
34
+ > **Explore → Adapt → Act**
35
+ > *Never assume. Always verify. Then execute.*
36
+
37
+ ```
38
+ ┌───────────────────────────────────────────────────────────┐
39
+ │ │
40
+ │ 🔍 EXPLORE → 🔄 ADAPT → ⚡ ACT │
41
+ │ │
42
+ │ Discover Adjust to Execute with │
43
+ │ the reality what you find confidence │
44
+ │ │
45
+ └───────────────────────────────────────────────────────────┘
46
+ ```
101
47
 
102
- ### 🚀 Core Capabilities
48
+ ### Agents & Their Philosophy
103
49
 
104
- | | Feature | What It Means |
105
- |:---:|:---|:---|
106
- | | **Multi-Session Parallel** | 50 isolated sessions running simultaneously. True multi-threading. |
107
- | 🔥 | **Parallel Execution** | Workers build different files at once. 10x faster. |
108
- | | **Non-Blocking Async** | Fire-and-forget. Commander never waits. Results collected automatically. |
109
- | 🧩 | **Smart Distribution** | One file = one worker. No conflicts. No stepping on each other. |
110
- | 🔗 | **Real-Time Sync** | Shared `.opencode/` state. All agents see updates instantly. |
111
- | 🛡️ | **Auto Verify & Test** | E2E tests, import checks, integration validation. Bugs get caught. |
112
- | 🩹 | **Self-Healing** | Auto-recovery from crashes, rate limits, context overflow. 3 retries. |
113
- | � | **Live Monitoring** | Track sessions, progress, queues. Toast notifications in real-time. |
50
+ | Agent | Role | Philosophy |
51
+ |:-----:|:-----|:-----------|
52
+ | 🎯 **Commander** | Orchestrator | Discover project Delegate intelligently → Loop until sealed |
53
+ | 📋 **Planner** | Strategist | Research docs Plan with parallelism → Document findings |
54
+ | 🔨 **Worker** | Implementer | Observe patterns Learn conventions → Implement with fit |
55
+ | **Reviewer** | Gatekeeper | Read standards Run tests Approve with evidence |
114
56
 
115
- ---
57
+ ### Guiding Principles
116
58
 
59
+ | Principle | Practice |
60
+ |:----------|:---------|
61
+ | 🔍 **Discover, Don't Assume** | Read `context.md`, detect tech stack, find build/test commands |
62
+ | 📚 **Evidence Over Memory** | Cite docs, run actual commands, complete only with proof |
63
+ | 🔄 **Adapt to the Project** | Match existing patterns, follow discovered conventions |
64
+ | 🤝 **Separation of Concerns** | Each agent does one thing excellently |
117
65
 
118
- ## 🏛️ How It Works
66
+ ---
119
67
 
120
- > **⚡ DELEGATE. PARALLELIZE. SEAL.** — One command unleashes full autonomous execution.
68
+ ## �️ Architecture
121
69
 
122
70
  ```
123
71
  /task "Build REST API"
@@ -147,17 +95,32 @@ Then in OpenCode:
147
95
  ♻️ LOOP 🎖️ SEALED
148
96
  ```
149
97
 
150
- ### 🔑 Key Systems
98
+ ---
151
99
 
152
- | System | What It Does |
153
- |:------:|:-------------|
154
- | **🔀 Multi-Session** | `delegate_task` → `client.session.create()` → new isolated session |
155
- | **⚡ Parallel Agents** | 50 sessions run simultaneously, `ConcurrencyController` manages slots |
156
- | **🖥️ Background Tasks** | `run_background` → async shell commands, non-blocking |
157
- | **🔄 Auto-Continuation** | `session.idle` → check TODOs → inject prompt → keep going |
158
- | **🛡️ Self-Healing** | Rate limits, crashes → 3 auto-retries per session |
100
+ ## Key Features
159
101
 
160
- > **Stop anytime:** `/stop`
102
+ | Feature | Description |
103
+ |:--------|:------------|
104
+ | ⚡ **50 Parallel Sessions** | True multi-threading with isolated contexts |
105
+ | 🔥 **Parallel File Builds** | Workers build different files simultaneously |
106
+ | 🧩 **Smart Distribution** | One file = one worker. No conflicts |
107
+ | � **Real-Time Sync** | Shared `.opencode/` state across all agents |
108
+ | �️ **Auto Verification** | E2E tests, import checks, integration validation |
109
+ | 🩹 **Self-Healing** | Auto-recovery with 3 retries per session |
110
+
111
+ ### Self-Healing Details
112
+
113
+ | Error Type | Recovery Action |
114
+ |:-----------|:----------------|
115
+ | Tool crash | Inject recovery prompt |
116
+ | Rate limit | Exponential backoff + retry |
117
+ | Context overflow | Smart compaction |
118
+ | Session timeout | Resume from checkpoint |
119
+ | Build failure | Loop back, fix, retry |
120
+
121
+ ---
122
+
123
+ ## 📸 Screenshots
161
124
 
162
125
  <div align="center">
163
126
  <p><strong>TUI</strong></p>
@@ -173,30 +136,13 @@ Then in OpenCode:
173
136
 
174
137
  ---
175
138
 
176
- ## 🔧 Self-Healing Features
177
-
178
- ```
179
- ┌────────────────────────────────────────────────────────────┐
180
- │ ERROR TYPE │ AUTO-RECOVERY ACTION │
181
- ├────────────────────────────────────────────────────────────┤
182
- │ Tool crash │ Inject recovery prompt │
183
- │ Rate limit │ Exponential backoff + retry │
184
- │ Context overflow │ Smart compaction │
185
- │ Thinking block error │ Auto-restructure │
186
- │ Session timeout │ Resume from checkpoint │
187
- │ Build failure │ Loop back, fix, retry │
188
- └────────────────────────────────────────────────────────────┘
189
- ```
190
-
191
- ---
192
-
193
139
  ## 📚 Documentation
194
140
 
195
141
  - **[System Architecture](docs/SYSTEM_ARCHITECTURE.md)** — Full technical deep-dive
196
142
 
197
143
  ---
198
144
 
199
- ## Uninstall
145
+ ## 🗑️ Uninstall
200
146
 
201
147
  ```bash
202
148
  npm uninstall -g opencode-orchestrator
@@ -204,9 +150,6 @@ npm uninstall -g opencode-orchestrator
204
150
 
205
151
  ---
206
152
 
207
- ## License
153
+ ## 📄 License
208
154
 
209
155
  MIT License. [LICENSE](LICENSE)
210
-
211
- ---
212
-
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": "0.9.24",
5
+ "version": "0.9.25",
6
6
  "author": "agnusdei1207",
7
7
  "license": "MIT",
8
8
  "repository": {