Data

Tools

Indexes

Applications

Experiments

Awesome

An open API service indexing awesome lists of open source software.

https://github.com/albertsikkema/claude-config-template

Reusable Claude Code configuration template with custom agents, slash commands, and structured documentation
https://github.com/albertsikkema/claude-config-template

agents ai automation claude claude-code configuration development-tools productivity template

Last synced: 7 months ago
JSON representation

Reusable Claude Code configuration template with custom agents, slash commands, and structured documentation

Awesome Lists containing this project

README 

          
# Claude Code Configuration Template



> **DEPRECATION NOTICE**

> 

> This repository is **no longer actively maintained**. I have switched to [OpenCode](https://opencode.ai/) - it's open source, feels faster, offers more model choices, and isn't tied to a single company. The Claude Code subscription limitations, rate limiting, and frequent "resources not available" errors also became frustrating.

> 

> This repository will remain available until **January 1, 2026** for reference purposes. Feel free to fork, copy, or use any code from this repository before then.

> 

> Updates will be sporadic at best.



---



A powerful, reusable configuration system for [Claude Code](https://claude.com/code) that supercharges your development workflow with custom agents, intelligent slash commands, and structured project documentation.



## ⚡ Quick Install



**First time installation or update (preserves your work):**

```bash

curl -fsSL https://raw.githubusercontent.com/albertsikkema/claude-config-template/main/install.sh | bash

```



**Install from a specific branch:**

```bash

curl -fsSL https://raw.githubusercontent.com/albertsikkema/claude-config-template/main/install.sh | bash -s -- --branch orchestrator-agent

```



**Clean reinstall (⚠️ overwrites important stuff):**

```bash

curl -fsSL https://raw.githubusercontent.com/albertsikkema/claude-config-template/main/install.sh | bash -s -- --force

```



**Clean reinstall from a specific branch:**

```bash

curl -fsSL https://raw.githubusercontent.com/albertsikkema/claude-config-template/main/install.sh | bash -s -- --branch orchestrator-agent --force

```



> **Note**: Always use `/main/install.sh` in the URL. The `--branch` argument specifies which branch's content to actually install.



**Optional: Start monitoring dashboard** (requires [Bun](https://bun.sh)):

```bash

./start-monitoring.sh

```

See the [Multi-Agent Observability Dashboard](#-multi-agent-observability-dashboard) section for details.



## 🗑️ Quick Uninstall



**Remove configuration (preserves thoughts/):**

```bash

curl -fsSL https://raw.githubusercontent.com/albertsikkema/claude-config-template/main/uninstall.sh | bash

```



**Remove everything including thoughts/ (⚠️ overwrites important stuff):**

```bash

curl -fsSL https://raw.githubusercontent.com/albertsikkema/claude-config-template/main/uninstall.sh | bash -s -- --force

```



---



## 🎯 What's This?



This is a **configuration template** that you install into your projects. It provides:



- **Complete development workflow** - Research → Plan → Implement → Cleanup → Deploy ([see WORKFLOW.md](WORKFLOW.md))

- **12 specialized AI agents** - Automated research, code analysis, and architecture design

- **14 slash commands** - Streamlined workflows for common tasks (including C4 architecture diagrams and deployment automation)

- **108 security rules** - Language-specific secure coding guidance from [Project Codeguard](https://github.com/project-codeguard/rules)

- **Multi-agent observability** - Real-time monitoring dashboard with AI-powered event summaries

- **Structured documentation system** - Templates and organization for project docs

- **Pre-configured permissions** - Ready-to-use tool access for development



Think of it as a **productivity multiplier** for Claude Code - install once, benefit forever.



**📖 New to this template? Start with [WORKFLOW.md](WORKFLOW.md) for the complete development workflow guide.**



Partly based on/ inspired by:

- https://github.com/humanlayer/humanlayer

- https://github.com/Wirasm/PRPs-agentic-eng



Security rules integration:

- https://github.com/project-codeguard/rules (108 OWASP-aligned security rules)



Monitoring dashboard:

- https://github.com/disler/claude-code-hooks-multi-agent-observability



## 📦 What You Get



### 🤖 Intelligent Agents



**Codebase Intelligence:**

- `codebase-locator` - Find WHERE code lives

- `codebase-analyzer` - Understand HOW code works

- `codebase-pattern-finder` - Discover similar implementations

- `codebase-researcher` - Orchestrate comprehensive research



**Architecture & Planning:**

- `system-architect` - Design systems and evaluate patterns

- `plan-implementer` - Execute approved technical plans



**Documentation Research:**

- `project-context-analyzer` - Extract and synthesize project documentation context

- `best-practices-researcher` - Search documented best practices from previous implementations

- `technical-docs-researcher` - Search technical documentation for libraries and frameworks

- `thoughts-analyzer` - Deep dive into your thoughts directory

- `thoughts-locator` - Find relevant documents



**External Research:**

- `web-search-researcher` - Research from the web



### ⚡ Slash Commands



| Command | Purpose |

|---------|---------|

| `/project` | Create project documentation from templates |

| `/research_codebase` | Deep codebase investigation |

| `/create_plan` | Interactive implementation planning |

| `/implement_plan` | Execute approved plans |

| `/validate_plan` | Validate implementation correctness |

| `/cleanup` | Document best practices and clean up ephemeral artifacts |

| `/build_c4_docs` | Generate C4 architecture diagrams (System Context, Container, Component) |

| `/commit` | Create well-formatted git commits |

| `/pr` | Generate comprehensive PR descriptions |

| `/code_reviewer` | Review code quality |

| `/security` | Comprehensive security analysis with Codeguard rules |

| `/deploy` | Automated deployment preparation (version, changelog, build, release) |

| `/fetch_technical_docs` | Fetch LLM-optimized documentation from context7.com |

| `/index_codebase` | Index Python/TypeScript/Go codebases |



### 📁 Directory Structure



After installation, you'll have:



```

your-project/

├── .claude/

│   ├── agents/              # 12 specialized agents

│   ├── commands/            # 14 slash commands

│   ├── hooks/               # Observability hooks (if monitoring enabled)

│   └── settings.json        # Configuration and hooks

│

├── docs/                    # Helper script documentation

│   ├── README-fetch-docs.md     # Documentation fetcher guide

│   ├── README-indexers.md       # Codebase indexers guide

│   ├── README-c4-diagrams.md    # C4 architecture diagrams guide

│   ├── README-fetch-openapi.md  # OpenAPI fetcher guide

│   └── README-spec-metadata.md  # Metadata generator guide

│

├── claude-helpers/          # Utility scripts

│   ├── README.md            # Scripts overview

│   ├── index_python.py      # Python codebase indexer

│   ├── index_js_ts.py       # JavaScript/TypeScript codebase indexer

│   ├── index_go.py          # Go codebase indexer

│   ├── build_c4_diagrams.py # C4 PlantUML diagram builder

│   ├── fetch-docs.py        # Documentation fetcher

│   ├── fetch_openapi.sh     # OpenAPI schema fetcher

│   └── spec_metadata.sh     # Metadata generator

│

└── thoughts/

    ├── templates/           # Documentation templates

    │   ├── project.md.template  # Project context template

    │   ├── todo.md.template     # Active work tracking template

    │   ├── done.md.template     # Completed work template

    │   ├── adr.md.template      # Architecture Decision Records template

    │   └── changelog.md.template # Changelog template

    │

    ├── best_practices/      # Best practices documentation from implementations

    ├── technical_docs/      # Technical documentation storage

    ├── security_rules/      # Project Codeguard security rules (108 rules)

    │   ├── core/            # 22 Cisco-curated core security rules

    │   └── owasp/           # 86 OWASP-based security rules

    │

    └── shared/

        ├── plans/           # Implementation plans (deleted after cleanup)

        ├── research/        # Research documents (deleted after cleanup)

        ├── rationalization/ # Ephemeral working docs (deleted after cleanup)

        └── project/         # Project documentation (3-file structure)

```



## 🔍 Multi-Agent Observability Dashboard



Gain real-time visibility into Claude Code's decision-making process with the monitoring dashboard. Watch agents spawn, tools execute, and see AI-generated summaries of all activities. Basically uses this repo: [https://github.com/disler/claude-code-hooks-multi-agent-observability](https://github.com/disler/claude-code-hooks-multi-agent-observability)



### What You Get



- **Real-time event streaming** - See every tool use, agent spawn, and notification as it happens

- **AI-powered summaries** - Automatically generated summaries of complex events

- **Session tracking** - Follow complete conversation flows and agent orchestration

- **Event filtering** - Focus on specific event types or sessions

- **WebSocket updates** - Live dashboard updates with no refresh needed



### Quick Start



**Prerequisites**: [Bun](https://bun.sh) must be installed

```bash

# Install Bun (macOS/Linux)

curl -fsSL https://bun.sh/install | bash

```



**Option 1: One-Line Setup** (Easiest)

```bash

./start-monitoring.sh

```



This will:

1. Clone the monitoring dashboard repository

2. Install all dependencies using Bun

3. Set up observability hooks in `.claude/hooks/`

4. Configure hooks in `.claude/settings.json`

5. Start the monitoring server and dashboard



**Option 2: Skip Hooks Setup** (if you have custom hooks)

```bash

./start-monitoring.sh --skip-hooks

```



**Option 3: Force Clean Reinstall**

```bash

./start-monitoring.sh --force-install

```



### Accessing the Dashboard



Once started, access:

- **Dashboard UI**: http://localhost:5173

- **Server API**: http://localhost:4000



### What Gets Tracked



The monitoring system captures:

- **PreToolUse** - Before any tool executes

- **PostToolUse** - After tool execution completes

- **Notification** - User input requests (with optional TTS)

- **Stop** - Conversation stop events

- **SubagentStop** - When spawned agents complete

- **PreCompact** - Before context window compaction

- **UserPromptSubmit** - User message submissions

- **SessionStart/End** - Session lifecycle events



### Helper Scripts



After starting, you can:

```bash

# Stop the monitoring system

./claude-code-hooks-multi-agent-observability/scripts/reset-system.sh



# Test the monitoring system

./claude-code-hooks-multi-agent-observability/scripts/test-system.sh

```



### How It Works



1. **Hooks** in `.claude/hooks/` intercept Claude Code events

2. **Events** are sent to the server at `http://localhost:4000`

3. **AI summaries** are generated for complex events (optional)

4. **Dashboard** displays everything in real-time via WebSocket



### Customization



The hooks configuration is in `.claude/settings.json`. You can:

- Disable specific hooks

- Adjust which events get AI summaries

- Change the server URL

- Add custom event processing



## 🚀 Installation Options



### Option 1: Use as GitHub Template 



1. Click **"Use this template"** on [GitHub](https://github.com/albertsikkema/claude-config-template)

2. Create your template repository

3. Clone and use in your projects



### Option 2: Direct Installation



```bash

# Clone the template

git clone https://github.com/albertsikkema/claude-config-template.git



# Navigate to your project

cd /path/to/your-project



# Install

/path/to/claude-config-template/install-helper.sh

```



### Option 3: One-Line Remote Install ⚡ (Recommended)



The easiest way! Downloads, installs, and cleans up automatically:



```bash

# Install everything (recommended)

curl -fsSL https://raw.githubusercontent.com/albertsikkema/claude-config-template/main/install.sh | bash



# Install only Claude configuration

curl -fsSL https://raw.githubusercontent.com/albertsikkema/claude-config-template/main/install.sh | bash -s -- --claude-only



# Install only thoughts structure

curl -fsSL https://raw.githubusercontent.com/albertsikkema/claude-config-template/main/install.sh | bash -s -- --thoughts-only



# Preview what will be installed (dry run)

curl -fsSL https://raw.githubusercontent.com/albertsikkema/claude-config-template/main/install.sh | bash -s -- --dry-run



# Install to specific directory

curl -fsSL https://raw.githubusercontent.com/albertsikkema/claude-config-template/main/install.sh | bash -s -- /path/to/project



# Combine options

curl -fsSL https://raw.githubusercontent.com/albertsikkema/claude-config-template/main/install.sh | bash -s -- --force --claude-only

```



**How it works**: The remote installer downloads the repository, runs the installation with your options, and automatically cleans up temporary files.



## 🎮 Installation Options



```bash

# Install everything (default)

./install-helper.sh



# Install only Claude configuration

./install-helper.sh --claude-only



# Install only thoughts structure

./install-helper.sh --thoughts-only



# Preview what will be installed

./install-helper.sh --dry-run



# Clean reinstall (⚠️ removes all thoughts/ content)

./install-helper.sh --force



# Install to specific directory

./install-helper.sh /path/to/project

```



**Note**:

- The remote installer (`install.sh`) downloads the repository, installs to your current directory, and cleans up automatically

- For manual installation from a cloned repository, use `install-helper.sh`

- The installer automatically updates your project's `.gitignore` to exclude `.claude/`, `thoughts/`, and `claude-helpers/`

- **Default behavior**: `.claude/` is always updated, `thoughts/` preserves existing content and adds missing directories

- **With `--force`**: Completely replaces `thoughts/` directory, removing all plans, research, and project docs



## 📚 Helper Scripts & Documentation



This template includes several utility scripts in the `claude-helpers/` directory:



- **Codebase Indexers**: Generate searchable markdown documentation

  - `index_python.py` - Index Python codebases (functions, classes, models)

  - `index_js_ts.py` - Index JavaScript/TypeScript/React codebases (components, functions, interfaces)

  - `index_go.py` - Index Go codebases (structs, interfaces, functions)

  - Use via `/index_codebase` slash command

  - **📖 See [docs/README-indexers.md](docs/README-indexers.md) for detailed guide**



- **Documentation Fetcher**: Download LLM-optimized documentation

  - `fetch-docs.py` - Fetch documentation from context7.com

  - Use via `/fetch_technical_docs` slash command

  - **📖 See [docs/README-fetch-docs.md](docs/README-fetch-docs.md) for detailed guide**



- **OpenAPI Fetcher**: Extract API schemas from FastAPI

  - `fetch_openapi.sh` - Fetch OpenAPI/Swagger schemas

  - Auto-invoked via `/index_codebase` when FastAPI detected

  - **📖 See [docs/README-fetch-openapi.md](docs/README-fetch-openapi.md) for detailed guide**



- **Metadata Generator**: Capture development context

  - `spec_metadata.sh` - Generate comprehensive metadata

  - Used in plans, research, and ADRs

  - **📖 See [docs/README-spec-metadata.md](docs/README-spec-metadata.md) for detailed guide**



- **Orchestrator Agent**: Automate the full Claude Code workflow

  - `orchestrator.py` - Runs index → research → plan → implement → review

  - Supports both OpenAI and Azure OpenAI (auto-detected from `.env.claude`)

  - Single-file script with `uv run` support

  ```bash

  # Create .env.claude with API key

  echo "OPENAI_API_KEY=sk-..." > .env.claude



  # Run full workflow

  uv run claude-helpers/orchestrator.py "Add user authentication"



  # Stop after planning (no implementation)

  uv run claude-helpers/orchestrator.py --no-implement "Refactor database"

  ```

  **Tip**: Add an alias for easy access:

  ```bash

  # Add to ~/.zshrc or ~/.bashrc

  alias orchestrate='uv run /path/to/claude-helpers/orchestrator.py'

  ```



**📖 Full scripts overview: [claude-helpers/README.md](claude-helpers/README.md)**



## 📚 Complete Development Workflow



This template provides a systematic **Research → Plan → Implement → Cleanup** workflow based on "Faking a Rational Design Process in the AI Era".



**📖 See [WORKFLOW.md](WORKFLOW.md) for the complete guide** covering:



- **Phase 0**: Index Codebase (optional but recommended)

- **Phase 1**: Project Setup (one-time)

- **Phase 2**: Research

- **Phase 3**: Plan

- **Phase 4**: Implement

- **Phase 5**: Validate

- **Phase 6**: Cleanup (MANDATORY - documents best practices, removes ephemeral artifacts)

- **Phase 7**: Commit & PR



### Quick Start Example



```bash

# 1. Index your codebase (optional but makes research faster)

You: /index_codebase



# 2. Document your project (one-time setup)

You: /project Create full docs for my e-commerce platform



# 3. Research before building

You: /research_codebase payment processing flow



# 4. Create implementation plan

You: /create_plan add Stripe payment integration



# 5. Implement the plan

You: /implement_plan thoughts/shared/plans/2025-10-14-stripe-integration.md



# 6. Validate implementation

You: /validate_plan thoughts/shared/plans/2025-10-14-stripe-integration.md



# 7. Cleanup (MANDATORY - documents best practices, removes artifacts)

You: /cleanup thoughts/shared/plans/2025-10-14-stripe-integration.md



# 8. Commit and create PR

You: /commit

You: /pr

```



**👉 Read [WORKFLOW.md](WORKFLOW.md) for detailed explanations, examples, and best practices.**



## 📝 File Naming Conventions



**Plans & Research**: `YYYY-MM-DD-description.md` or `YYYY-MM-DD-TICKET-123-description.md`

```

2025-10-14-oauth-support.md

2025-10-14-ENG-1478-user-tracking.md

```



**Project Documentation**: Ultra-lean 3-file structure

```

thoughts/shared/project/project.md    # Project context (what/why/stack)

thoughts/shared/project/todo.md       # Active work (Must Haves/Should Haves)

thoughts/shared/project/done.md       # Completed work history

```



**Best Practices**: Category-based naming as `[category]-[topic].md`

```

thoughts/best_practices/authentication-oauth-patterns.md

thoughts/best_practices/database-transaction-handling.md

thoughts/best_practices/api-error-handling.md

```



## 🎨 Customization



### Add Your Own Agent



Create `.claude/agents/my-agent.md`:



```markdown

---

name: my-custom-agent

description: What this agent does

model: sonnet

color: blue

---



Your agent instructions here...

```



### Add Custom Command



Create `.claude/commands/my-command.md`:



```markdown

# My Custom Command



Your command instructions here...

```



### Adjust Permissions



Edit `.claude/settings.json`:



```json

{

  "permissions": {

    "allow": [

      "Bash(command:*)",

      "WebFetch(domain:example.com)"

    ],

    "deny": []

  }

}

```



## 🗑️ Uninstallation



```bash

# Remove configuration (preserves thoughts/)

./uninstall.sh



# Remove only Claude config

./uninstall.sh --claude-only



# Preview what will be removed

./uninstall.sh --dry-run



# Remove everything including thoughts/ (⚠️ deletes your work)

./uninstall.sh --force

```



**Default behavior**: Removes `.claude/` and `claude-helpers/` but **preserves** `thoughts/` directory with your plans, research, and project docs.



**With `--force`**: Removes **everything** including all your work in `thoughts/`.



## 🌟 Key Features Explained



### Project Codeguard Security Integration



The `/security` command integrates **108 comprehensive security rules** from [Project Codeguard](https://github.com/project-codeguard/rules):



- **22 core rules** - Authentication, input validation, authorization, cryptography, etc.

- **86 OWASP rules** - SQL injection, XSS, CSRF, session management, etc.

- **Language-specific guidance** - Python, JavaScript, Go, Java, C, and more

- **Code examples** - Safe implementations and secure coding patterns

- **Automatic loading** - Rules filtered by your project's detected technology stack



When you run `/security`, it automatically:

1. Detects your languages and frameworks

2. Loads relevant Codeguard rules from `thoughts/security_rules/`

3. Analyzes your code against 18 security areas + Codeguard patterns

4. Generates reports with rule references and recommendations



**Source**: Curated by Cisco and aligned with OWASP best practices.



### Research → Plan → Implement → Cleanup Pattern



The core workflow ensures quality and preserves knowledge:



1. **Research**: Understand before building (spawns parallel agents, documents findings)

2. **Plan**: Design before coding (interactive planning with user)

3. **Implement**: Execute with clarity (step-by-step with validation)

4. **Cleanup**: Document best practices and remove ephemeral artifacts (see [WORKFLOW.md](WORKFLOW.md))



### Intelligent Agents



Agents work autonomously and can be:

- **Invoked automatically** by Claude Code when needed

- **Requested explicitly** by you

- **Run in parallel** for faster results



### Structured Documentation



Ultra-lean 3-file structure + best practices:

- **project.md** - Project context (what/why/stack/constraints)

- **todo.md** - Active work with MoSCoW prioritization (Must Haves/Should Haves)

- **done.md** - Completed work history with full traceability

- **best_practices/** - Documented patterns, decisions, and lessons learned from implementations



See the "Ultra-Lean 3-File Documentation Method" section in [WORKFLOW.md](WORKFLOW.md) for methodology details.



## 📖 Real-World Examples



### Example 1: New Feature Development



```

You: /research_codebase I need to add two-factor authentication



Claude uses agents to:

1. Research existing auth code (codebase-analyzer)

2. Find similar implementations (codebase-pattern-finder)

3. Check best practices (best-practices-researcher)

4. Check technical docs (technical-docs-researcher)



Then read and use the created research file to create a plan, implement etc.

```



### Example 2: Understanding Legacy Code



```

You: /research_codebase How does the caching layer work?



Claude spawns parallel agents to:

- Locate cache-related files

- Analyze implementation details

- Document the architecture



Result: Comprehensive research document with file:line references

```



### Example 3: Project Setup



```

You: /project Create complete project documentation



Claude asks about:

- Tech stack

- Team size

- Development phase

- Core features



Result: Full project documentation customized for your needs

```



## 🤝 Contributing



Contributions are welcome from individuals and companies alike! We'd love to see:



- New agents and slash commands

- Improved documentation and examples

- Bug fixes and enhancements

- Installation script improvements

- Templates and workflow ideas



**See [CONTRIBUTING.md](CONTRIBUTING.md) for detailed guidelines on:**

- How to submit contributions

- Agent and command development best practices

- Testing requirements

- Pull request process



For questions or larger contributions, contact: license@albertsikkema.com



## 💡 Tips & Best Practices



1. **Follow the workflow** - See [WORKFLOW.md](WORKFLOW.md) for the complete process

2. **Index first** - Run `/index_codebase` before research for faster results

3. **Research before planning** - Always understand before building

4. **Never skip cleanup** - It documents best practices and removes clutter for future AI sessions

5. **Document best practices** - Capture lessons learned, not just decisions

6. **Review permissions** - Audit `settings.local.json` regularly



## 📊 Version



Current version: **1.4.0**



## 🆘 Support & Troubleshooting



### Common Issues



**"Unknown slash command: project"**



If the `/project` command isn't recognized after installation:



1. **Restart Claude Code** - Close and reopen the application

2. **Check installation** - Verify `.claude/commands/project.md` exists

3. **Reload configuration** - Use the `/clear` command in Claude Code

4. **Re-install** - Run `./install-helper.sh --force` to ensure files are correct



**Agent not found**



If agents aren't being recognized:

- Verify `.claude/agents/` contains all 11 agent files

- Restart Claude Code

- Check file permissions (files should be readable)



**Git tracking configuration files**



The installer automatically adds `.claude/`, `thoughts/`, and `claude-helpers/` to your `.gitignore`. If you want to track these:

- Edit `.gitignore` and remove the entries you want to track

- Or use `git add -f` to force-add specific files



### Getting Help



- **Issues**: [GitHub Issues](https://github.com/albertsikkema/claude-config-template/issues)

- **Documentation**: [Claude Code Docs](https://docs.claude.com)

- **Template Updates**: Re-run the installer to update `.claude/` and add any new `thoughts/` directories



## 📄 License



MIT License - See LICENSE file for details



## 🎯 What Makes This Special?



Unlike basic Claude Code configurations, this template provides:



✅ **Complete workflow system** - Not just tools, but processes

✅ **Intelligent automation** - Agents that think and research

✅ **Structured knowledge** - Organized documentation system

✅ **Battle-tested patterns** - Proven workflows that work

✅ **Easy to customize** - Extend and adapt to your needs



## 🚦 Getting Started Checklist



After installation:



- [ ] **Restart Claude Code** to load new configuration

- [ ] Verify installation: Check `.claude/` and `thoughts/` directories exist

- [ ] Review `.claude/settings.json` permissions

- [ ] **(Optional)** Start monitoring dashboard: `./start-monitoring.sh`

- [ ] **Read [WORKFLOW.md](WORKFLOW.md)** to understand the complete process

- [ ] Run `/index_codebase` to create searchable indexes (optional but recommended)

- [ ] Run `/project` to document your project (if command not found, see troubleshooting)

- [ ] Follow the workflow: Research → Plan → Implement → Validate → Cleanup → PR

- [ ] Add project-specific agents if needed

- [ ] Share with your team!



---



**Ready to supercharge your development workflow?**



Install now and experience the future of AI-assisted development! 🚀



Made with ❤️ for the Claude Code community