pi-agent-setup

Full pi coding agent setup — extensions, skills, themes, agents, model router — ready for VPS deployment.

This repo mirrors the ~/.pi/ directory structure so you can clone it directly onto any machine and have your complete pi agent configuration ready to go.

Directory Structure

~/.pi/
├── agent/
│   ├── AGENTS.md                  # Global agent instructions
│   ├── settings.json              # pi settings (theme, provider, model)
│   ├── package.json               # Node dependencies (yaml parser)
│   ├── package-lock.json
│   ├── auth.json                  # API keys (empty placeholder — set your own)
│   ├── damage-control-rules.yaml  # Damage control rules for agents
│   │
│   ├── agents/                    # 12 agent definitions + teams.yaml + agent-chain.yaml
│   │   ├── bowser.md
│   │   ├── builder.md
│   │   ├── config-scout.md
│   │   ├── documenter.md
│   │   ├── pi-orchestrator.md
│   │   ├── planner.md
│   │   ├── plan-reviewer.md
│   │   ├── red-team.md
│   │   ├── researcher.md
│   │   ├── reviewer.md
│   │   ├── scout.md
│   │   ├── ultra-simple.md
│   │   ├── teams.yaml
│   │   └── agent-chain.yaml
│   │
│   ├── extensions/                # 7 active extensions
│   │   ├── cross-agent.ts
│   │   ├── minimal.ts
│   │   ├── session-replay.ts
│   │   ├── system-select.ts
│   │   ├── theme-cycler.ts
│   │   ├── tool-counter-widget.ts
│   │   └── wiki-session-docs.ts
│   │
│   ├── extensions-available/      # 8 optional extensions
│   │   ├── agent-chain.ts
│   │   ├── agent-team.ts
│   │   ├── pi-pi/                 # Pi-in-pi extension
│   │   ├── pi-pi.ts
│   │   ├── pure-focus.ts
│   │   ├── purpose-gate.ts
│   │   ├── subagent-widget.ts
│   │   ├── tilldone.ts
│   │   └── tool-counter.ts
│   │
│   ├── skills/                    # 6 agent skills
│   │   ├── brain-wiki/            # Brain knowledge base integration
│   │   ├── building-with-tambo/   # Tambo AI integration
│   │   ├── deep-research/         # Parallel.ai deep research
│   │   ├── find-skills/           # Skill discovery
│   │   ├── git-skills/            # Git operations
│   │   └── github-skills/         # GitHub operations
│   │
│   ├── themes/                    # 11 custom themes
│   │   ├── catppuccin-mocha.json
│   │   ├── cyberpunk.json
│   │   ├── dracula.json
│   │   ├── everforest.json
│   │   ├── gruvbox.json
│   │   ├── midnight-ocean.json
│   │   ├── nord.json
│   │   ├── ocean-breeze.json
│   │   ├── rose-pine.json
│   │   ├── synthwave.json
│   │   └── tokyo-night.json
│   │
│   ├── prompts/                   # 5 prompt templates
│   │   ├── ask.md
│   │   ├── brain.md
│   │   ├── ingest.md
│   │   ├── lint.md
│   │   └── research.md
│   │
│   ├── lib/                       # Shared libraries
│   │   └── themeMap.ts
│   │
│   └── docs/                      # Extension documentation
│       ├── COMPARISON.md
│       ├── PI_VS_OPEN_CODE.md
│       ├── RESERVED_KEYS.md
│       ├── THEME.md
│       └── TOOLS.md
│
├── model-router/                  # GLM Model Router (Python)
│   ├── model_router.py            # Core routing engine
│   ├── router_cli.py              # CLI interface
│   ├── router_server.py           # HTTP server
│   ├── test_router.py             # Tests
│   ├── examples.py                # Usage examples
│   ├── models-config.yaml         # Model configuration
│   ├── requirements.txt           # Python deps
│   ├── server-requirements.txt    # Server deps
│   ├── start.sh                   # Quick start
│   ├── README.md
│   ├── QUICKSTART.md
│   └── SUMMARY.md
│
├── Brain/                         # Brain Wiki (sync script only)
│   ├── sync.sh                    # Git sync script
│   ├── AGENTS.md                  # Wiki instructions
│   └── ONBOARDING.md              # Onboarding guide
│
└── examples/
    └── project-config/            # Project-level config templates
        ├── pi/
        │   ├── agents/            # 8 team agent definitions
        │   └── extensions/
        │       └── unibrain-head/ # Head agent extension
        └── agents/
            ├── teams/             # Team charters
            ├── pipeline/          # CI/CD pipeline config
            ├── config/            # Quality gates, constitution
            ├── agent-memory/      # Agent memory (empty)
            └── bug-reports/       # Bug reports (empty)

Quick Install (Fresh VPS)

Option 1: Using the install script

curl -fsSL https://raw.githubusercontent.com/JaskoCoder/pi-agent-setup/main/install.sh | bash

Option 2: Manual installation

# 1. Install Node.js 22
curl -fsSL https://deb.nodesource.com/setup_22.x | sudo -E bash -
sudo apt-get install -y nodejs

# 2. Install pi coding agent
npm install -g @mariozechner/pi-coding-agent

# 3. Install Python 3 (for model router)
sudo apt-get install -y python3 python3-pip

# 4. Clone this repo to ~/.pi
git clone https://github.com/JaskoCoder/pi-agent-setup.git ~/.pi

# 5. Install Node dependencies
cd ~/.pi/agent && npm install

# 6. Install model router dependencies
cd ~/.pi/model-router && pip3 install -r requirements.txt

# 7. Set up API keys
# Edit ~/.pi/agent/auth.json with your provider API keys, OR
# Set environment variables:
export OPENAI_API_KEY="sk-..."
export ANTHROPIC_API_KEY="sk-ant-..."
# etc.

# 8. Clone Brain Wiki (optional, requires access)
git clone https://github.com/JaskoCoder/brain-wiki.git ~/.pi/Brain

Using With a Project

To use this agent setup with a project, copy the project-level config:

# In your project root:
mkdir -p .pi .agents

# Copy agent definitions for your project teams
cp -r ~/.pi/examples/project-config/pi/agents/ .pi/agents/
cp -r ~/.pi/examples/project-config/pi/extensions/ .pi/extensions/

# Copy the autonomous organization
cp -r ~/.pi/examples/project-config/agents/* .agents/

Then customize:

• .pi/agents/*.md — Edit agent definitions for your project's tech stack
• .pi/extensions/unibrain-head/index.ts — Customize the head agent
• .agents/teams/*.md — Adjust team charters
• .agents/config/ — Set quality gates and pipeline config

Settings

The settings.json configures:

• Provider: zai-glm (GLM model provider)
• Model: glm-5
• Theme: synthwave (plus 10 others available)
• Damage Control: disabled

To change providers/models, edit ~/.pi/agent/settings.json or set them in pi's interactive config.

Themes

11 custom themes included:

• Catppuccin Mocha, Cyberpunk, Dracula, Everforest, Gruvbox
• Midnight Ocean, Nord, Ocean Breeze, Rosé Pine, Synthwave, Tokyo Night

Switch themes with the theme-cycler extension or edit settings.json.

Model Router

The GLM Model Router is a Python tool that routes tasks to the most appropriate model based on task characteristics. See model-router/README.md for full documentation.

Brain Wiki

The Brain Wiki is a private knowledge base that syncs via GitHub. To set it up:

# Clone the private wiki repo into ~/.pi/Brain
git clone https://github.com/JaskoCoder/brain-wiki.git ~/.pi/Brain

# Sync with the included script
bash ~/.pi/Brain/sync.sh --pull

Security

• auth.json is an empty placeholder — never commit API keys
• Brain Wiki content is excluded (private repo)
• No sessions or runtime state is included
• No binaries — tools like rg, fd are installed by pi

Requirements

• Node.js >= 22
• pi (@mariozechner/pi-coding-agent)
• Python >= 3.10 (for model router, optional)
• Git (for Brain Wiki sync)

License

Personal configuration — use freely as a template for your own pi setup.