Our take
Activeregular commits and releases- Activity
- 90
- Community
- 75
- Issues
- 95
- Pull requests
- 71
+6,543 stars in the last 30 days
- Releases more often than 93% of the projects we track
- Closes issues faster than 81% of the projects we track
- More contributors than 78% of the projects we track
Pi is an actively maintained, MIT-licensed TypeScript toolkit for developers who value a customizable coding-agent harness and can manage its permissions and Node.js requirement.
Good fit if
- You want an extensible coding-agent CLI, automation modes, or TypeScript components for an integration.
- You need to customize the workflow with extensions and packages instead of relying on built-in sub-agents or plan mode.
- You can use Node.js 22.19 or newer and are prepared to sandbox the agent if its default user-level permissions are too broad.
Look elsewhere if
- You require built-in sub-agents, plan mode, or a permission system that restricts filesystem, process, network, or credential access.
- Your environment cannot run Node.js 22.19 or newer.
- You need a ready-made agent workflow that cannot be adapted through extensions or packages.
All health signals
| Last commit | 2026-10-22 (today) |
|---|---|
| Commits, last 90 days | 100+ |
| Releases, last 12 months | 100+ (latest v1.4.0, 2026-10-21) |
| Contributors | 100+ (top contributor: 58% of commits) |
| Issues closed, last 90 days | 77+ (typically closed in 0 days) |
| Pull requests merged, last 90 days | 38 (typically merged in 0 days) |
| Project age | 15 months |
Checked on 2026-10-23 with the GitHub API.
Project history
Daily GitHub snapshots since Oct 7, 2026. History covers stars and forks; health signals are not tracked historically.
Stars
122,047Forks
15,734Change since Nov 14, 2026: +158 stars·+27 forks
Overview
Pi is a minimal, extensible agent harness built with TypeScript and the project's pi-ai unified LLM API. Its coding-agent CLI helps developers give an AI agent tasks in a project directory, while its supporting libraries provide reusable agent, terminal UI, and model API components.
Pi is designed to be adapted rather than impose a fixed workflow. It runs on macOS, Linux, and Windows, and can be used interactively, automated through print or JSON mode, controlled over RPC, or embedded through its TypeScript SDK. It deliberately leaves out built-in sub-agents and plan mode, so teams can add those through extensions or packages if needed.
Key Features
- Unified LLM API for providers including OpenAI, Anthropic, and Google.
- Interactive coding-agent CLI with tool calling and agent state management.
- Extensions, skills, prompt templates, themes, and shareable Pi packages for customization.
- Print and JSON modes for scripts and automation.
- RPC control and a TypeScript SDK for building integrations and applications.
- Supporting libraries for terminal UI, durable conversations and tasks, telemetry, and plugins.
- Installers for macOS, Linux, and Windows, plus Nix support for macOS and Linux.
Use Cases
- A developer wants an interactive AI coding assistant in a project directory and is comfortable granting it the permissions of the current user.
- A team wants to script coding-agent tasks and consume structured output through print or JSON mode.
- A developer building an application wants to use Pi's TypeScript SDK, agent runtime, or unified LLM API instead of adopting the CLI alone.
- A team wants to tailor agent behavior with custom skills, extensions, themes, or packages rather than use a fixed set of workflows.
What you need
Detected in the repository
- Node.js >=22.19.0 (from package.json)
- Automated checks on GitHub Actions
License in plain words
MITpermissive
- Commercial use: yes
- Modify and redistribute: yes
- You must keep: the copyright and license notice
- Share your changes: no
A summary, not legal advice: the LICENSE file is what applies.
Getting Started
Install the CLI with the official installer, then start it in the directory where you want it to work:
curl -fsSL https://pi.dev/install.sh | sh
cd /path/to/project
pi
Use /login to connect a provider subscription or API key. See the README and Pi documentation for setup options and usage details.
Alternatives
- kilocode: Kilo Code integrates with VS Code and JetBrains as well as the command line, and offers built-in modes for planning, debugging, and codebase questions.
- qwen-code: Qwen Code centers on Qwen models and offers interfaces beyond the terminal, including editor, desktop, browser, and chat integrations.
- opencode: OpenCode provides a terminal UI and beta desktop apps, with built-in build and read-only plan agents.
- gsd-pi: GSD Pi structures coding work into milestones and tasks, tracking project state and verification for longer-running jobs.
| Project | Language | License | Stars | Status |
|---|---|---|---|---|
| Pi | TypeScript | MIT | 117k | Active |
| kilocode | TypeScript | MIT | 27.5k | Active |
| qwen-code | TypeScript | Apache-2.0 | 28.6k | Active |
| opencode | TypeScript | MIT | 212k | Active |
| gsd-pi | TypeScript | MIT | 1.3k | New |
Considerations
- Pi does not provide built-in restrictions for filesystem, process, network, or credential access. It runs with the launching user's permissions; teams needing stronger boundaries should review the documented containerization options.
- The README explicitly omits built-in sub-agents and plan mode. Those need to be added through customization or a package if required.
- Node.js 22.19 or newer is required. The project offers installers for macOS, Linux, and Windows, while its documented Nix support covers macOS and Linux.
- The project is actively maintained, with regular commits and releases, and its recent activity involves at least 100 contributors. Contributor activity is not concentrated in one person; the measured issue-handling data also indicates issues and pull requests are being processed.
- The MIT license permits broad reuse, subject to its license terms and notice requirements.
Found this useful?
Share it with someone who would like Pi.