micode 0.9.0 → 0.10.0

package/README.md

@@ -79,7 +79,7 @@ Maintain context across sessions with structured compaction. Run `/ledger` to cr
 
 ## Hooks
 
-- **Think Mode** - Keywords like "think hard" enable 32k token thinking budget
+- **Think Mode** - Keywords like "think hard" enable 128k token thinking budget
 - **Ledger Loader** - Injects continuity ledger into system prompt
 - **Auto-Compact** - At 50% context usage, automatically summarizes session to reduce context
 - **File Ops Tracker** - Tracks read/write/edit for deterministic logging
@@ -106,11 +106,15 @@ All micode agents will use this model automatically.
 
 Create `~/.config/opencode/micode.json` for micode-specific settings:
 
-```json
+```jsonc
 {
   "agents": {
     "brainstormer": { "model": "openai/gpt-4o", "temperature": 0.8 },
-    "commander": { "maxTokens": 8192 }
+    "commander": {
+      "maxTokens": 8192,
+      // Configure reasoning effort per agent
+      "thinking": { "type": "enabled", "budgetTokens": 100000 }
+    }
   },
   "features": {
     "mindmodelInjection": true
@@ -122,11 +126,13 @@ Create `~/.config/opencode/micode.json` for micode-specific settings:
 }
 ```
 
+> **Note:** Both `.json` and `.jsonc` formats are supported. JSONC allows comments and trailing commas.
+
 #### Options
 
 | Option | Type | Description |
 |--------|------|-------------|
-| `agents` | object | Per-agent overrides (model, temperature, maxTokens) |
+| `agents` | object | Per-agent overrides (model, temperature, maxTokens, thinking) |
 | `features.mindmodelInjection` | boolean | Enable mindmodel context injection |
 | `compactionThreshold` | number | Context usage threshold (0-1) for auto-compaction. Default: 0.5 |
 | `fragments` | object | Additional prompt fragments per agent |
@@ -182,6 +188,36 @@ git push --follow-tags
 6. **Continuous verification** - Implementer + Reviewer per task
 7. **Session continuity** - Never lose context
 
+## micode vs oh-my-opencode
+
+Both are OpenCode plugins, but with different philosophies:
+
+| Aspect | micode | oh-my-opencode |
+|--------|--------|----------------|
+| **Philosophy** | Opinionated workflow (brainstorm→plan→implement) | Batteries-included framework |
+| **Agent Design** | Role-based (Brainstormer, Planner, Executor) | Greek mythology theme (Sisyphus, Atlas, Prometheus) |
+| **Parallelism** | Batch-first: 10-20 concurrent micro-tasks (2-5 min each) | Background tasks with tmux visual monitoring |
+| **Code Guidance** | Mindmodel system with project-specific patterns | Comment checker, keyword modes (ultrawork) |
+| **Context Recovery** | Ledger system (CONTINUITY files) | AGENTS.md hierarchy, preemptive compaction |
+| **Workflow** | TDD-enforced with adaptation over escalation | Category-based delegation (visual-engineering, ultrabrain) |
+| **Configuration** | Focused options | Extensive (34 hooks, 11 agents, fallback chains) |
+
+### When to Choose micode
+
+- You want a **structured brainstorm→plan→implement workflow**
+- You prefer **TDD-driven implementation** with test-first development
+- You need **project-specific pattern enforcement** via mindmodel
+- You want **high parallelism on granular tasks** (10-20 concurrent micro-tasks)
+- You value **session continuity** via structured ledgers
+
+### When to Choose oh-my-opencode
+
+- You want **maximum flexibility** and configuration options
+- You prefer **keyword-driven modes** (e.g., "ultrawork", "analyze")
+- You need **extensive model fallback chains** with subscription detection
+- You like **category-based task delegation** (visual-engineering, infrastructure)
+- You want **visual monitoring** via tmux integration
+
 ## Inspiration
 
 - [oh-my-opencode](https://github.com/code-yeongyu/oh-my-opencode) - Plugin architecture