mirror of
https://github.com/catlog22/Claude-Code-Workflow.git
synced 2026-02-12 02:37:45 +08:00
Enhance workflows and commands for intelligent tools strategy
- Updated intelligent-tools-strategy.md to include `--skip-git-repo-check` for Codex write access and development commands. - Improved context gathering and analysis processes in mcp-tool-strategy.md with additional examples and guidelines for file searching. - Introduced new command concept-enhanced.md for enhanced intelligent analysis with parallel CLI execution and design blueprint generation. - Added context-gather.md command for intelligent collection of project context based on task descriptions, generating standardized JSON context packages.
This commit is contained in:
catlog22
@@ -1,188 +0,0 @@
|
||||
---
|
||||
name: auto
|
||||
description: Auto-select and execute appropriate template based on user input analysis
|
||||
usage: /gemini:mode:auto "description of task or problem"
|
||||
argument-hint: "description of what you want to analyze or plan"
|
||||
examples:
|
||||
- /gemini:mode:auto "authentication system keeps crashing during login"
|
||||
- /gemini:mode:auto "design a real-time notification architecture"
|
||||
- /gemini:mode:auto "database connection errors in production"
|
||||
- /gemini:mode:auto "plan user dashboard with analytics features"
|
||||
allowed-tools: Bash(ls:*), Bash(gemini:*)
|
||||
model: sonnet
|
||||
---
|
||||
|
||||
# Auto Template Selection (/gemini:mode:auto)
|
||||
|
||||
## Overview
|
||||
Automatically analyzes user input to select the most appropriate template and execute Gemini CLI with optimal context.
|
||||
|
||||
**Directory Analysis Rule**: Intelligent detection of directory context intent - automatically navigate to target directory when analysis scope is directory-specific.
|
||||
|
||||
**--cd Parameter Rule**: When `--cd` parameter is provided, always execute `cd "[path]" && gemini --all-files -p "prompt"` to ensure analysis occurs in the specified directory context.
|
||||
|
||||
**Process**: List Templates → Analyze Input → Select Template → Execute with Context
|
||||
|
||||
## Usage
|
||||
|
||||
### Auto-Detection Examples
|
||||
```bash
|
||||
# Bug-related keywords → selects bug-fix.md
|
||||
/gemini:mode:auto "React component not rendering after state update"
|
||||
|
||||
# Planning keywords → selects plan.md
|
||||
/gemini:mode:auto "design microservices architecture for user management"
|
||||
|
||||
# Error/crash keywords → selects bug-fix.md
|
||||
/gemini:mode:auto "API timeout errors in production environment"
|
||||
|
||||
# Architecture/design keywords → selects plan.md
|
||||
/gemini:mode:auto "implement real-time chat system architecture"
|
||||
|
||||
# With directory context
|
||||
/gemini:mode:auto "authentication issues" --cd "src/auth"
|
||||
```
|
||||
|
||||
## Template Selection Logic
|
||||
|
||||
### Dynamic Template Discovery
|
||||
**Templates auto-discovered from**: `~/.claude/prompt-templates/`
|
||||
|
||||
Templates are dynamically read from the directory, including their metadata (name, description, keywords) from the YAML frontmatter.
|
||||
|
||||
### Template Metadata Parsing
|
||||
|
||||
Each template contains YAML frontmatter with:
|
||||
```yaml
|
||||
---
|
||||
name: template-name
|
||||
description: Template purpose description
|
||||
category: template-category
|
||||
keywords: [keyword1, keyword2, keyword3]
|
||||
---
|
||||
```
|
||||
|
||||
**Auto-selection based on:**
|
||||
- **Template keywords**: Matches user input against template-defined keywords
|
||||
- **Template name**: Direct name matching (e.g., "bug-fix" matches bug-related queries)
|
||||
- **Template description**: Semantic matching against description text
|
||||
|
||||
## Command Execution
|
||||
|
||||
### Step 1: Template Discovery
|
||||
```bash
|
||||
# Dynamically discover all templates and extract YAML frontmatter
|
||||
cd "~/.claude/prompt-templates" && echo "Discovering templates..." && for template_file in *.md; do echo "=== $template_file ==="; head -6 "$template_file" 2>/dev/null || echo "Error reading $template_file"; echo; done
|
||||
```
|
||||
|
||||
### Step 2: Dynamic Template Analysis & Selection
|
||||
```pseudo
|
||||
FUNCTION select_template(user_input):
|
||||
templates = list_directory("~/.claude/prompt-templates/")
|
||||
template_metadata = {}
|
||||
|
||||
# Parse all templates for metadata
|
||||
FOR each template_file in templates:
|
||||
content = read_file(template_file)
|
||||
yaml_front = extract_yaml_frontmatter(content)
|
||||
template_metadata[template_file] = {
|
||||
"name": yaml_front.name,
|
||||
"description": yaml_front.description,
|
||||
"keywords": yaml_front.keywords || [],
|
||||
"category": yaml_front.category || "general"
|
||||
}
|
||||
|
||||
input_lower = user_input.toLowerCase()
|
||||
best_match = null
|
||||
highest_score = 0
|
||||
|
||||
# Score each template against user input
|
||||
FOR each template, metadata in template_metadata:
|
||||
score = 0
|
||||
|
||||
# Keyword matching (highest weight)
|
||||
FOR each keyword in metadata.keywords:
|
||||
IF input_lower.contains(keyword.toLowerCase()):
|
||||
score += 3
|
||||
|
||||
# Template name matching
|
||||
IF input_lower.contains(metadata.name.toLowerCase()):
|
||||
score += 2
|
||||
|
||||
# Description semantic matching
|
||||
FOR each word in metadata.description.split():
|
||||
IF input_lower.contains(word.toLowerCase()) AND word.length > 3:
|
||||
score += 1
|
||||
|
||||
IF score > highest_score:
|
||||
highest_score = score
|
||||
best_match = template
|
||||
|
||||
# Default to first template if no matches
|
||||
RETURN best_match || templates[0]
|
||||
END FUNCTION
|
||||
```
|
||||
|
||||
### Step 3: Execute with Dynamically Selected Template
|
||||
```bash
|
||||
# Basic execution with selected template
|
||||
gemini --all-files -p "$(cat ~/.claude/prompt-templates/[selected_template])
|
||||
|
||||
User Input: [user_input]"
|
||||
|
||||
# With --cd parameter
|
||||
cd "[specified_directory]" && gemini --all-files -p "$(cat ~/.claude/prompt-templates/[selected_template])
|
||||
|
||||
User Input: [user_input]"
|
||||
```
|
||||
|
||||
**Template selection is completely dynamic** - any new templates added to the directory will be automatically discovered and available for selection based on their YAML frontmatter.
|
||||
|
||||
|
||||
### Manual Template Override
|
||||
```bash
|
||||
# Force specific template
|
||||
/gemini:mode:auto "user authentication" --template bug-fix.md
|
||||
/gemini:mode:auto "fix login issues" --template plan.md
|
||||
```
|
||||
|
||||
### Dynamic Template Listing
|
||||
```bash
|
||||
# List all dynamically discovered templates
|
||||
/gemini:mode:auto --list-templates
|
||||
# Output:
|
||||
# Dynamically discovered templates in ~/.claude/prompt-templates/:
|
||||
# - bug-fix.md (用于定位bug并提供修改建议) [Keywords: 规划, bug, 修改方案]
|
||||
# - plan.md (软件架构规划和技术实现计划分析模板) [Keywords: 规划, 架构, 实现计划, 技术设计, 修改方案]
|
||||
# - [any-new-template].md (Auto-discovered description) [Keywords: auto-parsed]
|
||||
```
|
||||
|
||||
**Complete template discovery** - new templates are automatically detected and their metadata parsed from YAML frontmatter.
|
||||
|
||||
## Auto-Selection Examples
|
||||
|
||||
### Dynamic Selection Examples
|
||||
```bash
|
||||
# Selection based on template keywords and metadata
|
||||
"login system crashes on startup" → Matches template with keywords: [bug, 修改方案]
|
||||
"design user dashboard with analytics" → Matches template with keywords: [规划, 架构, 技术设计]
|
||||
"database timeout errors in production" → Matches template with keywords: [bug, 修改方案]
|
||||
"implement real-time notification system" → Matches template with keywords: [规划, 实现计划, 技术设计]
|
||||
|
||||
# Any new templates added will be automatically matched
|
||||
"[user input]" → Dynamically matches against all template keywords and descriptions
|
||||
```
|
||||
|
||||
|
||||
## Session Integration
|
||||
|
||||
saves to:
|
||||
`.workflow/WFS-[topic]/.chat/auto-[template]-[timestamp].md`
|
||||
|
||||
**Session includes:**
|
||||
- Original user input
|
||||
- Template selection reasoning
|
||||
- Template used
|
||||
- Complete analysis results
|
||||
|
||||
This command streamlines template usage by automatically detecting user intent and selecting the optimal template for analysis.
|
||||
@@ -1,140 +0,0 @@
|
||||
---
|
||||
name: plan-precise
|
||||
description: Precise path planning analysis for complex projects
|
||||
usage: /gemini:mode:plan-precise "planning topic"
|
||||
examples:
|
||||
- /gemini:mode:plan-precise "design authentication system"
|
||||
- /gemini:mode:plan-precise "refactor database layer architecture"
|
||||
---
|
||||
|
||||
### 🚀 Command Overview: `/gemini:mode:plan-precise`
|
||||
|
||||
Precise path-based planning analysis using user-specified directories instead of --all-files.
|
||||
|
||||
### 📝 Execution Template
|
||||
|
||||
```pseudo
|
||||
# Precise path planning with user-specified scope
|
||||
|
||||
PLANNING_TOPIC = user_argument
|
||||
PATHS_FILE = "./planning-paths.txt"
|
||||
|
||||
# Step 1: Check paths file exists
|
||||
IF not file_exists(PATHS_FILE):
|
||||
Write(PATHS_FILE, template_content)
|
||||
echo "📝 Created planning-paths.txt in project root"
|
||||
echo "Please edit file and add paths to analyze"
|
||||
# USER_INPUT: User edits planning-paths.txt and presses Enter
|
||||
wait_for_user_input()
|
||||
ELSE:
|
||||
echo "📁 Using existing planning-paths.txt"
|
||||
echo "Current paths preview:"
|
||||
Bash(grep -v '^#' "$PATHS_FILE" | grep -v '^$' | head -5)
|
||||
# USER_INPUT: User confirms y/n
|
||||
user_confirm = prompt("Continue with these paths? (y/n): ")
|
||||
IF user_confirm != "y":
|
||||
echo "Please edit planning-paths.txt and retry"
|
||||
exit
|
||||
|
||||
# Step 2: Read and validate paths
|
||||
paths_ref = Bash(.claude/scripts/read-paths.sh "$PATHS_FILE")
|
||||
|
||||
IF paths_ref is empty:
|
||||
echo "❌ No valid paths found in planning-paths.txt"
|
||||
echo "Please add at least one path and retry"
|
||||
exit
|
||||
|
||||
echo "🎯 Analysis paths: $paths_ref"
|
||||
echo "📋 Planning topic: $PLANNING_TOPIC"
|
||||
|
||||
# BASH_EXECUTION_STOPS → MODEL_ANALYSIS_BEGINS
|
||||
```
|
||||
|
||||
### 🧠 Model Analysis Phase
|
||||
|
||||
After bash script prepares paths, model takes control to:
|
||||
|
||||
1. **Present Configuration**: Show user the detected paths and analysis scope
|
||||
2. **Request Confirmation**: Wait for explicit user approval
|
||||
3. **Execute Analysis**: Run gemini with precise path references
|
||||
|
||||
### 📋 Execution Flow
|
||||
|
||||
```pseudo
|
||||
# Step 1: Present plan to user
|
||||
PRESENT_PLAN:
|
||||
📋 Precise Path Planning Configuration:
|
||||
|
||||
Topic: design authentication system
|
||||
Paths: src/auth/**/* src/middleware/auth* tests/auth/**/* config/auth.json
|
||||
Gemini Reference: $(.claude/scripts/read-paths.sh ./planning-paths.txt)
|
||||
|
||||
⚠️ Continue with analysis? (y/n)
|
||||
|
||||
# Step 2: MANDATORY user confirmation
|
||||
IF user_confirms():
|
||||
# Step 3: Execute gemini analysis
|
||||
Bash(gemini -p "$(.claude/scripts/read-paths.sh ./planning-paths.txt) @{CLAUDE.md} $(cat ~/.claude/prompt-templates/plan.md)
|
||||
|
||||
Planning Topic: $PLANNING_TOPIC")
|
||||
ELSE:
|
||||
abort_execution()
|
||||
echo "Edit planning-paths.txt and retry"
|
||||
```
|
||||
|
||||
### ✨ Features
|
||||
|
||||
- **Root Level Config**: `./planning-paths.txt` in project root (no subdirectories)
|
||||
- **Simple Workflow**: Check file → Present plan → Confirm → Execute
|
||||
- **Path Focused**: Only analyzes user-specified paths, not entire project
|
||||
- **No Complexity**: No validation, suggestions, or result saving - just core function
|
||||
- **Template Creation**: Auto-creates template file if missing
|
||||
|
||||
### 📚 Usage Examples
|
||||
|
||||
```bash
|
||||
# Create analysis for authentication system
|
||||
/gemini:mode:plan-precise "design authentication system"
|
||||
|
||||
# System creates planning-paths.txt (if needed)
|
||||
# User edits: src/auth/**/* tests/auth/**/* config/auth.json
|
||||
# System confirms paths and executes analysis
|
||||
```
|
||||
|
||||
### 🔍 Complete Execution Example
|
||||
|
||||
```bash
|
||||
# 1. Command execution
|
||||
$ /gemini:mode:plan-precise "design authentication system"
|
||||
|
||||
# 2. System output
|
||||
📋 Precise Path Planning Configuration:
|
||||
|
||||
Topic: design authentication system
|
||||
Paths: src/auth/**/* src/middleware/auth* tests/auth/**/* config/auth.json
|
||||
Gemini Reference: @{src/auth/**/*,src/middleware/auth*,tests/auth/**/*,config/auth.json}
|
||||
|
||||
⚠️ Continue with analysis? (y/n)
|
||||
|
||||
# 3. User confirms
|
||||
$ y
|
||||
|
||||
# 4. Actual gemini command executed
|
||||
$ gemini -p "$(.claude/scripts/read-paths.sh ./planning-paths.txt) @{CLAUDE.md} $(cat ~/.claude/prompt-templates/plan.md)
|
||||
|
||||
Planning Topic: design authentication system"
|
||||
```
|
||||
|
||||
### 🔧 Path File Format
|
||||
|
||||
Simple text file in project root: `./planning-paths.txt`
|
||||
|
||||
```
|
||||
# Comments start with #
|
||||
src/auth/**/*
|
||||
src/middleware/auth*
|
||||
tests/auth/**/*
|
||||
config/auth.json
|
||||
docs/auth/*.md
|
||||
```
|
||||
|
||||
Reference in New Issue
Block a user