There are many coding agents, but this one is mine.

$ npm install -g @mariozechner/pi-coding-agent

Minimal. Opinionated. Extensible.

Lean core keeps your context clean. Use only what you need.

Core

  • 4 toolsread, write, edit, bash
  • Multi-providerAnthropic, OpenAI, Google, Mistral, Groq, Cerebras, xAI, OpenRouter, ZAI, MiniMax, Amazon Bedrock, Ollama, plus OAuth for Copilot, Gemini CLI, Antigravity, Claude, Codex
  • /modelswitch models mid-conversation
  • Sessionstree view, branching, resume any point
  • Compactionauto-summarize long conversations
  • Bash modeprefix with ! to run commands
  • Imagespaste or drag into context
  • Model cyclingCtrl+P/Shift+Ctrl+P, --models patterns
  • Message queuingsteering vs follow-up messages

Opt-in

What we didn't build

pi is opinionated about what it won't do. These are intentional design decisions to minimize context bloat and avoid anti-patterns.

No MCP. Build CLI tools with READMEs (see Skills). The agent reads them on demand. Would you like to know more? If you must, use mcporter.

No sub-agents. Spawn pi instances via tmux, or build your own sub-agent extension. Observability and steerability for you and the agent.

No permission popups. Security theater. Run in a container or build your own with Extensions.

No plan mode. Gather context in one session, write plans to file, start fresh for implementation.

No built-in to-dos. They confuse models. Use a TODO.md file, or build your own with extensions.

No background bash. Use tmux. Full observability, direct interaction.

Read the blog post for the full rationale.

Get involved

Bug reports and feature requests: Please use GitHub Issues. This is the only channel actively monitored by maintainers.

Discord: There's an unofficial community Discord for general discussion. Note that maintainers do not actively monitor this channel.

MIT License • Mario Zechner & contributors