Production-ready system prompts for Claude AI and coding agents. Built on the APEI cycle: Analyze → Plan → Execute → Iterate.
🤖 Agent Prompts (Autonomous AI Agents)
| Prompt |
Purpose |
Tokens |
File |
| Agent System ⭐ |
Core operating system |
~1.5K |
View |
| Error Analysis |
Debugging & root cause |
~2K |
View |
| Project Workflow |
Full lifecycle management |
~2.5K |
View |
| Quick Reference |
Cheat sheet (minimal) |
~0.8K |
View |
| Code Review |
Systematic PR review |
~2K |
View |
| Security Audit |
Vulnerability detection |
~2.5K |
View |
| Refactoring |
Code improvement |
~2.5K |
View |
| Testing |
Test design & TDD |
~3K |
View |
| Documentation |
Technical writing |
~2.5K |
View |
| Performance |
Optimization & profiling |
~3K |
View |
| Git & VCS |
Branching & commits |
~2.5K |
View |
| Integration Guardian |
System integrity |
~0.5K |
View |
🎯 Claude Code-Specific Prompts
| Prompt |
Purpose |
Tokens |
File |
| Mode Transitions ⭐ |
/think, /ultrathink, /compact modes & planning |
~2.5K |
View |
| Token Optimization |
Token saving strategies for Claude Code |
~2.5K |
View |
| Workflow & Config |
CLAUDE.md, hooks, permissions, MCP |
~3K |
View |
🏗️ Architecture & Full-Stack Prompts
| Prompt |
Purpose |
Tokens |
File |
| Technology Stack ⭐ |
Library discovery, modern tools, hidden gems |
~4K |
View |
| Architecture Patterns |
System design, microservices, DDD, CQRS |
~3.5K |
View |
| Full-Stack Development |
Next.js/Nuxt/SvelteKit, end-to-end type safety |
~3.5K |
View |
| Prompt Chaining |
Multi-step workflows, context management |
~3K |
View |
📋 Foundation & Project Prompts (Interactive Sessions)
| Prompt |
Purpose |
Technologies |
File |
| Foundation ⭐ |
Universal best practices |
Any |
View |
| Web Development |
Frontend apps |
React, Vue, Angular, CSS |
View |
| API Development |
Backend services |
REST, GraphQL, Node, Go |
View |
| Data Science & ML |
ML pipelines |
Python, pandas, PyTorch |
View |
| Mobile |
Mobile apps |
iOS, Android, Flutter, RN |
View |
| DevOps & CI/CD |
Infrastructure |
K8s, Docker, Terraform |
View |
| Database & SQL |
Data layer |
PostgreSQL, MySQL, Redis |
View |
| General Software |
Cross-language |
Python, JS, Go, Java, C# |
View |
| Language |
Status |
Prompts |
Link |
| English |
✅ Complete |
19 agent + 8 project |
Index |
| Türkçe |
🔄 In Progress |
7 prompts |
Index |
For AI Agents (Claude Code, Copilot)
1. Copy content of claude-agent-system-prompt.md
2. Paste as system prompt
3. Give task → Agent runs APEI cycle automatically
1. Copy Foundation prompt + project-type prompt
2. Paste at start of Claude session
3. Describe your task → Follow the APEI cycle together
| Context Budget |
Recommended Setup |
| < 2K tokens |
Quick Reference only |
| 2K–8K tokens |
Agent System Prompt |
| 8K+ tokens |
Agent System + project-type prompt |
| Project Type |
Prompts to Use |
| AI Agent Task |
Agent System |
| Claude Code Setup |
Agent System + Workflow & Config |
| Claude Code Planning |
Agent System + Mode Transitions |
| Token-Efficient Session |
Agent System + Token Optimization |
| Debug / Fix Bugs |
Agent System + Error Analysis |
| New Project |
Agent System + Project Workflow |
| Choose Best Tools |
Agent System + Technology Stack |
| System Architecture |
Agent System + Architecture Patterns |
| Full-Stack App (modern) |
Agent System + Full-Stack Development |
| Complex Multi-Step Task |
Agent System + Prompt Chaining |
| Code Review |
Agent System + Code Review |
| React / Vue App |
Foundation + Web Development |
| REST API |
Foundation + API Development |
| ML Model |
Foundation + Data Science & ML |
| Mobile App |
Foundation + Mobile |
| DevOps / Infra |
Foundation + DevOps & CI/CD |
prompts/
├── english/
│ ├── agents/ # Agent-optimized prompts (19 files)
│ ├── base/ # Foundation prompt
│ ├── project-types/ # Domain-specific prompts (7 files)
│ ├── examples/ # Real-world usage examples
│ └── workflows/ # APEI methodology guide
├── turkish/
│ ├── agents/ # Türkçe agent prompts (6 files)
│ ├── base/ # Türkçe foundation prompt
│ └── INDEX.md
Fork → Modify prompts → Add team standards → Use.
All prompts are MIT-licensed templates. Adapt for your stack, conventions, and tools.
See CONTRIBUTING.md. Key areas: new languages, new project types, prompt improvements, examples.
MIT License — use, modify, and distribute freely.