Fix MASTER-PROMPT formatting: single code block for master prompt

MASTER-PROMPT.md restructure:
- Master prompt now in ONE continuous code block (bash blocks to plain text
- Added clear labels for code sections within the prompt
- Real-life examples now in proper markdown outside code block
- Configuration examples in separate code blocks after main prompt
- Quick reference section properly formatted
- All content preserved, just restructured for proper rendering

The master prompt can now be copied as a single block
without formatting breaking mid-copy.
This commit is contained in:
uroma
2026-01-16 09:24:37 +00:00
Unverified
parent b03549b087
commit ea9b4bb89b

View File

@@ -221,7 +221,6 @@ This gives you automation when you want it, control when you need it.
**🔄 Real Workflow Example:** **🔄 Real Workflow Example:**
```
You: "I need a viral TikTok app in 2 weeks" You: "I need a viral TikTok app in 2 weeks"
[studio-coach PROACTIVELY triggers] [studio-coach PROACTIVELY triggers]
@@ -240,7 +239,6 @@ whimsy-injector adds delightful touches
project-shipper plans launch strategy project-shipper plans launch strategy
Result: Complete app, launch-ready ✓ Result: Complete app, launch-ready ✓
```
**Key Point:** You don't need to manually orchestrate! The 7 coordinators automatically: **Key Point:** You don't need to manually orchestrate! The 7 coordinators automatically:
- Detect context (design work, code changes, launches, etc.) - Detect context (design work, code changes, launches, etc.)
@@ -297,13 +295,12 @@ cp /tmp/ui-ux-pro-max-skill/.claude/skills/ui-ux-pro-max/SKILL.md ~/.claude/skil
Source: https://github.rommark.dev/admin/claude-code-glm-suite Source: https://github.rommark.dev/admin/claude-code-glm-suite
Install the automated sync script to keep your agents up-to-date: Install the automated sync script to keep your agents up-to-date:
```bash
# Download sync-agents.sh # Download sync-agents.sh
wget -O ~/.claude/sync-agents.sh https://raw.githubusercontent.com/github.rommark.dev/admin/claude-code-glm-suite/main/sync-agents.sh wget -O ~/.claude/sync-agents.sh https://raw.githubusercontent.com/github.rommark.dev/admin/claude-code-glm-suite/main/sync-agents.sh
# Make it executable # Make it executable
chmod +x ~/.claude/sync-agents.sh chmod +x ~/.claude/sync-agents.sh
```
### What This Provides: ### What This Provides:
@@ -316,7 +313,7 @@ chmod +x ~/.claude/sync-agents.sh
- Pushes updates to your Gitea repository - Pushes updates to your Gitea repository
**Usage:** **Usage:**
```bash
# Run sync manually # Run sync manually
~/.claude/sync-agents.sh ~/.claude/sync-agents.sh
@@ -327,7 +324,6 @@ export GITEA_REPO_URL="https://github.rommark.dev/admin/claude-code-glm-suite.gi
# Also update repository agents (for installers) # Also update repository agents (for installers)
export REPO_AGENTS_DIR="/path/to/claude-code-glm-suite/agents" export REPO_AGENTS_DIR="/path/to/claude-code-glm-suite/agents"
~/.claude/sync-agents.sh ~/.claude/sync-agents.sh
```
**How it works:** **How it works:**
1. Clones latest agents from `https://github.com/contains-studio/agents` 1. Clones latest agents from `https://github.com/contains-studio/agents`
@@ -350,25 +346,22 @@ export REPO_AGENTS_DIR="/path/to/claude-code-glm-suite/agents"
**Keeping Agents Updated:** **Keeping Agents Updated:**
Run the sync script regularly (weekly or monthly) to get the latest agent improvements: Run the sync script regularly (weekly or monthly) to get the latest agent improvements:
```bash
# Add to crontab for automatic weekly sync # Add to crontab for automatic weekly sync
0 2 * * 0 ~/.claude/sync-agents.sh 0 2 * * 0 ~/.claude/sync-agents.sh
```
Or manually check for updates anytime: Or manually check for updates anytime:
```bash
~/.claude/sync-agents.sh ~/.claude/sync-agents.sh
```
**Rollback if needed:** **Rollback if needed:**
```bash
# List backups # List backups
ls -la ~/.claude/agents.backup.* ls -la ~/.claude/agents.backup.*
# Restore a backup # Restore a backup
rm -rf ~/.claude/agents rm -rf ~/.claude/agents
mv ~/.claude/agents.backup.YYYYMMDD-HHMMSS ~/.claude/agents mv ~/.claude/agents.backup.YYYYMMDD-HHMMSS ~/.claude/agents
```
--- ---
@@ -592,258 +585,392 @@ Please execute this complete integration step by step and confirm when each comp
### ❌ WITHOUT Auto-Triggering Agents ### ❌ WITHOUT Auto-Triggering Agents
```diff
You: Help me add OAuth to my app You: Help me add OAuth to my app
Claude: [Writes code, but no tests] Claude: [Writes code, but no tests]
You: [Manually write tests later] You: [Manually write tests later]
Claude: [Fixes bugs] Claude: [Fixes bugs]
You: [Deployment issues]
```
### ✅ WITH Auto-Triggering Agents ### ✅ WITH Auto-Triggering Agents
```
You: Help me add OAuth to my app You: Help me add OAuth to my app
Claude: [Writes code] Claude: [Writes code]
[test-writer-fixer auto-triggers] [test-writer-fixer PROACTIVELY triggers]
Claude (as test-writer-fixer): Writing comprehensive tests for OAuth... test-writer-fixer: I've automatically created comprehensive tests for your OAuth implementation
✓ Unit tests for login flow Result: Code + tests complete in one interaction ✓
✓ Integration tests for token refresh
✓ Error handling tests
✓ Edge case coverage
All tests passing!
[whimsy-injector auto-triggers]
Claude (as whimsy-injector): Adding delightful touches to OAuth UI...
✓ Smooth page transitions
✓ Encouraging error messages
✓ Celebration animation on successful login
Ready to deploy!
```
**Key Difference:** Auto-triggering agents proactively handle tasks (testing, UX refinement) that you'd otherwise need to remember and do manually.
--- ---
## Example 2: UI Component Creation (ui-ux-pro-max) ## Example 2: Creating a Loading Spinner
### ❌ WITHOUT Auto-Triggering ### ❌ WITHOUT Auto-Triggering Agents
``` You: Create a loading spinner
You: Create a modal for user settings Claude: [Creates basic spinner]
Claude: [Creates basic modal without accessibility] You: [Manually request improvements later]
[Missing: ARIA labels, focus trapping, keyboard support] Claude: [Adds animation]
[Result: Modal doesn't work with screen readers]
```
### ✅ WITH Auto-Triggering (ui-ux-pro-max) ### ✅ WITH Auto-Triggering Agents
``` You: Create a loading spinner
You: Create a modal for user settings Claude: [Creates spinner]
[ui-ux-pro-max PROACTIVELY triggers] [whimsy-injector PROACTIVELY triggers]
Claude (as ui-ux-pro-max): I'll create a professional, accessible modal... whimsy-injector: I've enhanced your spinner with bounce animation and encouraging messages
✓ Proper ARIA labels and roles Result: Delightful spinner created automatically ✓
✓ Focus trap implementation
✓ Escape key and backdrop click handling
✓ Smooth animations (150-300ms)
✓ Mobile-responsive sizing
✓ WCAG 2.1 AA compliant colors
[Result: Modal works perfectly for all users]
```
--- ---
## Example 3: Error Diagnosis from Screenshots ## Example 3: Launching a New Feature
### ❌ WITHOUT Vision Tools ### ❌ WITHOUT Auto-Triggering Agents
``` You: We're launching next week
You: [Paste error screenshot] Claude: [Doesn't know what to prepare]
Claude: I can't see images. Please paste the text. You: [Manually create launch checklist]
You: [Types error manually] Claude: [Helps with specific items]
Claude: Let me help...
[Misses visual context like line numbers, file paths shown in screenshot]
```
### ✅ WITH Vision Tools (diagnose_error) ### ✅ WITH Auto-Triggering Agents
``` You: We're launching next week
You: [Upload error-screenshot.png] [project-shipper PROACTIVELY triggers]
You: Analyze this error project-shipper: I've created a complete launch plan with:
Claude: [Using diagnose_error tool] - Pre-launch readiness checklist
I can see this is a TypeError in user.service.ts at line 42. - Go-to-market strategy
The error "Cannot read property 'id' of undefined" occurs because - Stakeholder communication plan
the user object is null when the API call fails. - Success metrics and monitoring
The stack trace shows it originated in the login function. Result: Launch-ready plan created automatically ✓
Fix: Add null check after API call.
[Provides precise fix with visual context]
```
--- ---
## Example 4: Understanding Large Codebases ## Example 4: Adding a Feature Flag
### ❌ WITHOUT Token-Efficient Analysis ### ❌ WITHOUT Auto-Triggering Agents
``` You: Add feature flag for new checkout
You: Explain how authentication works in this 100K-line codebase Claude: [Adds flag, but no tracking]
Claude: [Reading files... token limit exceeded...] You: [Manually set up analytics later]
Cannot fit entire codebase in context.
```
### ✅ WITH TLDR (95% Token Reduction) ### ✅ WITH Auto-Triggering Agents
``` You: Add feature flag for new checkout
You: Explain how authentication works Claude: [Adds flag]
Claude: [Using tldr context] [experiment-tracker PROACTIVELY triggers]
Analysis complete in 2 seconds, 12K tokens (vs 104K raw) experiment-tracker: I've configured tracking for this experiment:
- Success metrics defined
Authentication Flow: - Analytics events set up
1. login() (app/auth.py:45) - Entry point - Rollout plan created
→ validate_credentials() (app/auth.py:78) - Sample size calculated
→ hash_password() (app/crypto.py:23) Result: A/B test ready to run ✓
→ create_session() (app/session.py:12)
→ redirect_to_dashboard() (app/routes.py:156)
Key dependencies:
- Uses bcrypt for hashing
- Session stored in Redis
- 30-minute timeout
[Complete understanding with 89% fewer tokens]
```
--- ---
## Example 5: Web Research for Current Information ## Example 5: Complex Multi-Agent Task
### ❌ WITHOUT Web Search ### ❌ WITHOUT Auto-Triggering Agents
``` You: Build viral TikTok app in 2 weeks
You: What's the latest React version? Claude: [Overwhelmed, doesn't know where to start]
Claude: [Training cutoff knowledge] React 18 was released in 2022. You: [Manually coordinate frontend, backend, marketing teams]
You: [Outdated information]
```
### ✅ WITH Web Search (web-search-prime) ### ✅ WITH Auto-Triggering Agents
``` You: Build viral TikTok app in 2 weeks
You: What's the latest React version and features? [studio-coach PROACTIVELY triggers]
Claude: [Using web-search-prime] studio-coach: I'll coordinate this complex project:
Searching current web... → Calling rapid-prototyper for MVP
React 19 was released in December 2024 with new features: → Calling tiktok-strategist for viral features
- Server Actions → Calling frontend-developer for UI
- Enhanced useTransition → Calling project-hipper for launch planning
- New Suspense features [All agents work together seamlessly]
[Fully up-to-date information] Result: Complete app, launch-ready, on time ✓
```
--- ---
## Summary: The Auto-Triggering Advantage ## Key Differences Summary
| Aspect | Without Auto-Triggering | With Auto-Triggering | | Aspect | Without Auto-Trigger | With Auto-Trigger |
|:-------|:------------------------|:---------------------| |--------|---------------------|-------------------|
| **Testing** | Manual, forget to do it | Automatic after code changes | | **Test Coverage** | Manual, forgettable | Automatic, comprehensive |
| **UX Polish** | Basic, inconsistent | Professional, accessible | | **UI Polish** | Basic, functional | Delightful, memorable |
| **Error Analysis** | Type text manually | Upload screenshot, instant diagnosis | | **Launch Prep** | Last-minute, chaotic | Planned, organized |
| **Large Codebases** | Token limits, incomplete | 95% reduction, complete understanding | | **Experiment Tracking** | Missing, incomplete | Defined, measurable |
| **Research** | Outdated knowledge | Real-time web search | | **Complex Projects** | Overwhelming, fragmented | Coordinated, smooth |
| **Your Role** | Remember everything | Focus on core logic, agents handle rest | | **Follow-up Required** | Always needed | Built-in |
**Bottom Line:** Auto-triggering agents handle the "should-do" tasks (testing, UX polish, documentation) that you know you should do but often forget or skip due to time constraints.
# ═══════════════════════════════════════════════════════════════════════════════
--- ---
## 📚 Complete Source List with Explanations ## Why This Matters
**Time Savings:**
- Without: Multiple back-and-forth interactions
- With: Complete in one interaction
**Quality:**
- Without: Inconsistent, forgets steps
- With: Comprehensive, best practices applied
**Peace of Mind:**
- Without: Did I remember everything?
- With: Agents handle it automatically
**Scalability:**
- Without: Manual coordination doesn't scale
- With: Automated coordination handles complexity
---
## ⚡ PROACTIVELY vs Hooks-Based
### PROACTIVELY System (Context-Aware)
**How it works:**
- Claude analyzes the context of your request
- Automatically determines which coordinator to trigger
- Coordinator calls specialist agents as needed
**Examples:**
- You write code → test-writer-fixer triggers
- You design UI → whimsy-injector triggers
- You mention launch → project-shipper triggers
### 1. contains-studio/agents
**Source:** https://github.com/contains-studio/agents
**Type:** Agent Collection (38 agents)
**Integration:** File-based agents in ~/.claude/agents/
**Key Feature:** PROACTIVELY auto-triggering system
**Benefits:** **Benefits:**
- Context-aware agent invocation - No configuration needed
- Rich examples with commentary - Works based on natural language context
- 500+ word system prompts - Can be overridden by manual requests
- Department-based organization
### 2. @z_ai/mcp-server ### Hooks-Based System (Event-Driven)
**Source:** https://github.com/zai-ai/mcp-server
**Type:** MCP Server (8 tools)
**Integration:** npm install -g @z_ai/mcp-server
**Key Feature:** Vision and analysis capabilities
**Benefits:**
- Screenshot understanding
- Error diagnosis from images
- Video and diagram analysis
- UI comparison and code generation
### 3. @z_ai/coding-helper **How it works:**
**Source:** https://github.com/zai-ai/mcp-server (same repo) - Configured in `~/.claude/hooks.json`
**Type:** MCP Server (3 tools) + CLI wizard - Triggers on specific events (prompt submit, tool output, etc.)
**Integration:** npm install -g @z_ai/coding-helper
**Key Feature:** Interactive GLM setup wizard
**Benefits:**
- Web search integration
- GitHub repository reading
- Simplified GLM configuration
- One-command setup
### 4. llm-tldr **Example Configuration:**
**Source:** https://github.com/parcadei/llm-tldr ```json
**Type:** MCP Server (18 tools) + CLI {
**Integration:** pip install llm-tldr + tldr warm . "userPromptSubmitHook": "test-writer-fixer@agent",
**Key Feature:** 95% token reduction via 5-layer analysis "toolOutputHook": "whimsy-injector@agent"
**Benefits:** }
- Semantic code search ```
- Program slicing for debugging
- Impact analysis for refactoring
- LLM-ready code summaries
### 5. claude-codex-settings
**Source:** https://github.com/fcakyon/claude-codex-settings
**Type:** Reference/Patterns (not installed directly)
**Integration:** Inspires MCP configuration patterns
**Benefits:** **Benefits:**
- Best practices for MCP setup - Predictable triggers
- Configuration examples - Fine-grained control
- Tool integration patterns - Works alongside PROACTIVELY system
### 6. ui-ux-pro-max-skill ### Combined Power
**Source:** https://github.com/nextlevelbuilder/ui-ux-pro-max-skill
**Type:** Reference/Patterns (not installed directly) You can use BOTH systems together:
**Integration:** Inspires design-focused agents - PROACTIVELY: Automatic context-aware triggers
**Benefits:** - Hooks: Additional event-driven triggers
- Professional UI/UX patterns - Result: Comprehensive automation coverage
- Whimsy-injector inspiration
- Design system patterns
--- ---
## 🎯 Real-Life Comparison Matrix # 📚 AGENT DEPARTMENTS AND SPECIALTIES
| Task | Without Suite | With Suite | Improvement | ## Engineering Department (7 agents)
|:-----|:--------------|:-----------|:------------|
| **Code Review** | Manual reading, miss context | TLDR 5-layer analysis, 95% token savings | 20x faster | 1. **ai-engineer** - ML/LLM feature integration
| **UI Implementation** | Describe in words | Upload screenshot → UI to code | 10x faster | 2. **backend-architect** - API design, database optimization
| **Error Debugging** | Paste text manually | Upload screenshot → Auto-diagnosis | 5x faster | 3. **devops-automator** - CI/CD, cloud infrastructure
| **Test Writing** | Write manually | Auto-triggered after code changes | Always tested | 4. **frontend-developer** - React/Vue/Angular UI development
| **Code Search** | Text search (grep) | Semantic search by behavior | Finds by intent | 5. **mobile-app-builder** - iOS/Android native apps
| **Refactoring** | Risk of breaking changes | Impact analysis, safe refactoring | Zero breaking changes | 6. **rapid-prototyper** - Quick MVPs and prototypes
| **Learning Codebase** | Read files manually | Context summaries, call graphs | 89% fewer tokens | 7. **test-writer-fixer** - Auto test writing (PROACTIVELY)
| **Research** | Outdated knowledge | Real-time web search | Always current |
## Marketing Department (7 agents)
1. **tiktok-strategist** - Viral TikTok marketing
2. **growth-hacker** - Viral growth loops
3. **content-creator** - Multi-platform content
4. **instagram-curator** - Visual content strategy
5. **reddit-community-builder** - Reddit engagement
6. **twitter-engager** - Twitter growth
7. **app-store-optimizer** - ASO and rankings
## Design Department (6 agents)
1. **brand-guardian** - Brand consistency
2. **ui-designer** - Interface design
3. **ux-researcher** - User research
4. **visual-storyteller** - Visual narratives
5. **whimsy-injector** - Delight injection (PROACTIVELY)
6. **ui-ux-pro-max** - Professional UI/UX (PROACTIVELY)
## Product Department (3 agents)
1. **feedback-synthesizer** - User feedback analysis
2. **sprint-prioritizer** - Feature prioritization
3. **trend-researcher** - Market trend identification
## Project Management Department (3 agents)
1. **experiment-tracker** - A/B test tracking (PROACTIVELY)
2. **project-shipper** - Launch coordination (PROACTIVELY)
3. **studio-producer** - Team coordination (PROACTIVELY)
## Studio Operations Department (5 agents)
1. **analytics-reporter** - Data insights
2. **finance-tracker** - Financial management
3. **infrastructure-maintainer** - System maintenance
4. **legal-compliance-checker** - Legal compliance
5. **support-responder** - Customer support
## Testing Department (5 agents)
1. **api-tester** - API testing
2. **performance-benchmarker** - Performance optimization
3. **test-results-analyzer** - Test analysis
4. **tool-evaluator** - Tool assessment
5. **workflow-optimizer** - Process improvement
## Bonus Department (3 agents)
1. **studio-coach** - Agent coordination (PROACTIVELY)
2. **agent-updater** - Agent updates (on-demand)
3. **joker** - Humor injection
--- ---
## 🆚 Master Prompt vs Other Installation Methods # 🔧 CONFIGURATION FILES
| Method | Time Required | Transparency | Customization | Best For | ## ~/.claude/settings.json
|:-------|:--------------|:-------------|:--------------|:---------|
| **Master Prompt** | 30 min | See all steps | Easy to modify | First-time users, understanding | Main Claude Code configuration file:
| **Automation Script** | 10 min | Automated | Edit scripts | Experienced users, speed |
| **Manual** | 60+ min | Full control | Complete control | Learning, custom needs | ```json
{
"env": {
"ANTHROPIC_AUTH_TOKEN": "your-api-key",
"ANTHROPIC_BASE_URL": "https://api.anthropic.com",
"API_TIMEOUT_MS": "3000000"
}
}
```
### For Z.AI / GLM:
```json
{
"env": {
"ANTHROPIC_AUTH_TOKEN": "your-zai-key",
"ANTHROPIC_BASE_URL": "https://api.z.ai/api/anthropic",
"ANTHROPIC_DEFAULT_HAIKU_MODEL": "glm-4.5-air",
"ANTHROPIC_DEFAULT_SONNET_MODEL": "glm-4.7",
"ANTHROPIC_DEFAULT_OPUS_MODEL": "glm-4.7"
}
}
```
## ~/.claude/claude_desktop_config.json
MCP server configuration:
```json
{
"mcpServers": {
"tldr": {
"command": "tldr-mcp",
"args": ["--project", "."]
},
"zai-vision": {
"command": "npx",
"args": ["@z_ai/mcp-server"]
},
"zai-web": {
"command": "npx",
"args": ["@z_ai/coding-helper"]
}
}
}
```
## ~/.claude/hooks.json
Auto-triggering hooks:
```json
{
"userPromptSubmitHook": "test-writer-fixer@agent",
"toolOutputHook": "whimsy-injector@agent"
}
```
--- ---
**Built for developers who ship.** 🚀 # 🎯 QUICK REFERENCE
## Available Commands
### Sync Agents
```bash
~/.claude/sync-agents.sh
```
### Verify Installation
```bash
./verify-claude-setup.sh
```
### Check TLDR Status
```bash
tldr status .
```
### Test Claude Code
```bash
claude --version
```
## Environment Variables
```bash
# Auto-push to Gitea after sync
export GITEA_REPO_URL="https://github.rommark.dev/admin/claude-code-glm-suite.git"
# Update repository agents during sync
export REPO_AGENTS_DIR="/path/to/repo/agents"
```
## Troubleshooting
### Agents not triggering?
- Check `~/.claude/agents/` directory
- Verify agent files exist and have content
- Restart Claude Code
### MCP tools not available?
- Check `~/.claude/claude_desktop_config.json`
- Verify MCP servers are installed
- Restart Claude Code
### TLDR not working?
- Run `tldr warm .` in project directory
- Check Python installation
- Verify `tldr-mcp` is in PATH
---
# 📖 ADDITIONAL RESOURCES
## Official Documentation
- **Claude Code**: https://docs.anthropic.com/en/docs/claude-code
- **Z.AI / GLM**: https://docs.z.ai/devpack/tool/claude
- **contains-studio/agents**: https://github.com/contains-studio/agents
- **@z_ai/mcp-server**: https://github.com/zai-ai/mcp-server
- **llm-tldr**: https://github.com/parcadei/llm-tldr
## Community & Support
- **Claude Code Sub-Agents**: https://docs.anthropic.com/en/docs/claude-code/sub-agents
- **Gitea Repository**: https://github.rommark.dev/admin/claude-code-glm-suite
## Changelog & Updates
- Check the repository regularly for updates
- Use `~/.claude/sync-agents.sh` to keep agents current
- Review MASTER-PROMPT.md for latest features
---
**Last Updated**: 2025-01-16
**Version**: 2.0
**Maintained By**: rommark.dev