oh-my-customcode 0.8.0 → 0.9.0

package/templates/.codex/rules/MUST-intent-transparency.md

@@ -0,0 +1,199 @@
+# [MUST] Intent Transparency Rules
+
+> **Priority**: MUST - Required for user control
+> **ID**: R015
+
+## Purpose
+
+Ensure transparency when automatically detecting user intent and routing to agents. Users should always understand why a specific agent was chosen.
+
+## Detection Display (REQUIRED)
+
+When intent is detected, display the reasoning:
+
+```
+[Intent Detected]
+├── Input: "{user input}"
+├── Agent: {detected-agent}
+├── Confidence: {percentage}%
+└── Reason: {explanation}
+```
+
+## Confidence Thresholds
+
+| Confidence | Action |
+|------------|--------|
+| >= 90% | Auto-execute with display |
+| 70-89% | Request confirmation |
+| < 70% | List options for user to choose |
+
+## Detection Factors
+
+Intent is detected using multiple factors:
+
+```yaml
+factors:
+  keywords:
+    weight: 40
+    description: Language-specific keywords in input
+    example: "Go", "Python", "리뷰", "생성"
+
+  file_patterns:
+    weight: 30
+    description: File extensions mentioned or in context
+    example: "*.go", "main.py"
+
+  action_verbs:
+    weight: 20
+    description: Action words in user input
+    example: "review", "create", "fix", "리뷰", "생성"
+
+  context:
+    weight: 10
+    description: Recent conversation context
+    example: Previous agent used, current working directory
+```
+
+## High Confidence Display (>= 90%)
+
+```
+[Intent Detected]
+├── Input: "Go 코드 리뷰해줘"
+├── Agent: lang-golang-expert
+├── Confidence: 95%
+└── Reason: "Go" keyword (40%) + "리뷰" action verb (55%)
+
+Executing...
+```
+
+## Medium Confidence Display (70-89%)
+
+```
+[Intent Detected]
+├── Input: "백엔드 코드 체크해줘"
+├── Agent: be-go-backend-expert (?)
+├── Confidence: 75%
+└── Reason: "백엔드" keyword (40%) + "체크" action (35%)
+
+Possible alternatives:
+  1. be-go-backend-expert (75%)
+  2. be-fastapi-expert (65%)
+  3. be-springboot-expert (60%)
+
+Proceed with be-go-backend-expert? [Y/n/1-3]
+```
+
+## Low Confidence Display (< 70%)
+
+```
+[Intent Unclear]
+├── Input: "이 코드 좀 봐줘"
+├── Confidence: < 70%
+└── Need more context
+
+Available agents for code review:
+  1. lang-golang-expert (Go files)
+  2. lang-python-expert (Python files)
+  3. lang-typescript-expert (TypeScript files)
+  4. lang-rust-expert (Rust files)
+
+Which agent should review the code? [1-4]
+```
+
+## Override Syntax
+
+Users can explicitly specify an agent:
+
+```
+@{agent-name} {command}
+```
+
+Examples:
+```
+@lang-golang-expert review this code
+@lang-python-expert src/main.py 분석해줘
+@secretary create a new agent
+```
+
+Override bypasses intent detection:
+
+```
+[Override] Agent explicitly specified: lang-python-expert
+Executing...
+```
+
+## Implementation
+
+### Secretary Workflow
+
+```
+1. Receive user input
+2. Check for explicit override (@agent)
+3. If no override:
+   a. Extract keywords, file patterns, action verbs
+   b. Match against agent triggers
+   c. Calculate confidence score
+   d. Display detection reasoning
+   e. Execute or request confirmation based on threshold
+4. Route to selected agent
+```
+
+### Detection Log
+
+For debugging and transparency:
+
+```yaml
+detection_log:
+  input: "Go 코드 리뷰해줘"
+  extracted:
+    keywords: ["Go"]
+    file_patterns: []
+    action_verbs: ["리뷰"]
+  matches:
+    - agent: lang-golang-expert
+      score: 95
+      breakdown:
+        keyword_go: 40
+        action_review: 40
+        context: 15
+  selected: lang-golang-expert
+  confidence: 95%
+```
+
+## Agent Triggers
+
+Each agent has defined triggers in:
+`.codex/skills/intent-detection/patterns/agent-triggers.yaml`
+
+```yaml
+agents:
+  lang-golang-expert:
+    keywords: [go, golang, "go 언어"]
+    file_patterns: ["*.go", "go.mod", "go.sum"]
+    actions: [review, analyze, fix, optimize]
+    base_confidence: 40
+
+  lang-python-expert:
+    keywords: [python, py, "파이썬"]
+    file_patterns: ["*.py", "requirements.txt", "pyproject.toml"]
+    actions: [review, analyze, fix, optimize]
+    base_confidence: 40
+```
+
+## Benefits
+
+1. **Transparency**: Users understand agent selection
+2. **Control**: Users can override or choose alternatives
+3. **Trust**: Clear reasoning builds user confidence
+4. **Learning**: Users learn which agents handle what
+5. **Debugging**: Clear logs for troubleshooting
+
+## Violations
+
+Proceeding without displaying intent reasoning = Rule violation
+
+The user must always know:
+- Which agent was selected
+- Why it was selected
+- What confidence level
+- How to override if needed

package/templates/.codex/rules/MUST-language-policy.md

@@ -0,0 +1,62 @@
+# [MUST] Language & Delegation Policy
+
+> **Priority**: MUST - Never violate
+> **ID**: R000
+
+## Core Principles
+
+### 1. User Input Language
+```
+User prompts are in Korean.
+Always understand and process Korean input.
+```
+
+### 2. Output Language
+
+| Context | Language | Example |
+|---------|----------|---------|
+| User communication | Korean | 상태 보고, 질문, 설명 |
+| Code | English | variables, functions, comments |
+| File contents | English | .md, .yaml, configs |
+| Commit messages | English | git commits |
+| Error messages to user | Korean | 에러 설명, 해결 방안 |
+
+### 3. Delegation Model
+```
+User does NOT directly edit files.
+User delegates ALL file operations to AI agent.
+
+User → (Korean prompt) → Agent → (file operations)
+```
+
+### 4. Context Efficiency
+```
+All file contents in English for:
+- Token efficiency
+- Consistent parsing
+- Universal compatibility
+```
+
+## Examples
+
+### Correct
+```
+User: "새로운 에이전트를 만들어줘"
+Agent: "에이전트를 생성하겠습니다." (Korean to user)
+Agent writes AGENT.md in English
+```
+
+### Incorrect
+```
+Agent writes: "# 에이전트 이름" in file  ← Wrong
+Agent writes: "# Agent Name" in file     ← Correct
+```
+
+## File Naming
+
+| Type | Convention | Example |
+|------|------------|---------|
+| Rules | `{PRIORITY}-{name}.md` | `MUST-safety.md` |
+| Agents | `AGENT.md` | - |
+| Skills | `SKILL.md` | - |
+| Index | `index.yaml` | - |