pi_agent_rust
High-performance AI coding agent CLI written in Rust with zero unsafe code
The Lens
By Erik Loyd, SaaS CEO and former COO/CFO of an AWS Premier Partner.
Updated Aug 2026
A from-scratch Rust port of Mario Zechner's Pi Agent, done with his blessing. Single binary called `pi`, no Node runtime and no Electron shell, 28 built-in tools with 18 live in a default session, and streaming that holds up over a long conversation. The complaint it answers is startup time and memory, and it answers it by rewriting the engine rather than trimming the old one.
Security is designed in rather than bolted on. Extension execution is capability gated across tool, exec, http, session, ui, and events, then a second stage mediates the command itself and blocks recursive delete, raw disk writes, and reverse shells by default. Extensions move through a trust lifecycle from pending to trusted to killed, with audit logs on the kill switch. More thought than most agent CLIs have given the problem.
Install is a curl script, and the only recurring cost is the model API key you already have.
The catch is the license, and it is a real one. GitHub's badge says MIT. The LICENSE file is MIT plus a rider that grants no rights whatsoever to OpenAI, Anthropic, or their affiliates, and forbids use in any training corpus, evaluation harness, or ML pipeline. Breach terminates the license automatically. That is not open source under the OSI definition.
Free vs Self-Hosted vs Paid
source availableFree (Source Available)
The binary and the source cost nothing and the install script is public. Every tool, the extension system, and the security model are in the repo.
The License Caveat
Read it before adopting. The file is MIT with an appended rider: OpenAI, Anthropic, and any entity they control or that controls them are Restricted Parties with no granted rights, and no one may make the software or a derivative available to them. The rider defines "use" to include benchmarking, testing, analyzing, indexing, or incorporating the code into any dataset, training corpus, evaluation harness, or ML pipeline. Any breach terminates permissions immediately, and the rider must be passed through unmodified on redistribution.
That fails the OSI definition on discrimination against persons and fields of endeavor. Call it source available, not open source. If your employer scans dependency licenses, this will trip the scan.
Self-Hosted Costs
Nothing beyond your machine. It is a native binary with no runtime, which is the whole point.
The Real Cost
Model API tokens, exactly as with any terminal coding agent. Bring an Anthropic, OpenAI, or compatible key. The rider restricts who may hold rights in the software; it does not stop you calling those vendors' APIs from it.
Free to download and run with no paid tier, but the MIT license carries a rider denying all rights to OpenAI, Anthropic, and their affiliates. Source available, not open source.
What to do by team size
- Solo
- free; the license rider is unlikely to touch you
- Small team
- free, but read the rider if you contract for a major AI lab
- Medium team
- free; flag it with legal before it lands in a dependency manifest
- Large team
- Expect a license scan to reject it; get an exception or use an OSI-licensed agent CLI
Get tools like this every Wednesday
One featured tool, three on the radar. No fluff.
Similar Tools
an open source, extensible AI agent that goes beyond code suggestions - install, execute, edit, and test with any LLM
The open source coding agent.
An open-source AI coding agent that lives in your terminal.
⌥ AI Coding agent for the terminal: hash-anchored edits, optimized tool harness, LSP, Python, browser, subagents, and more
Autonomous coding agent right in your IDE, capable of creating/editing files, executing commands, using the browser, and more with your permission every step of the way.
The free coding agent
A low score is not a verdict on quality. Young and niche tools start low by design. How we calculate scores
Trust Signals
License: MIT License (with rider)
MIT plus a rider granting no rights to OpenAI, Anthropic, or their affiliates, and barring use in any training corpus or evaluation harness. Fails the OSI definition.
Commercial use: ✓ Yes
About
- Owner
- Jeff Emanuel (User)
- Stars
- 1,689
- Forks
- 201
Explore Further
More tools in the directory
everything-claude-code
The agent harness performance optimization system. Skills, instincts, memory, security, and research-first development for Claude Code, Codex, Opencode, Cursor and beyond.
245.9k ★hermes-agent
The agent that grows with you
239.6k ★cc-switch
A cross-platform desktop All-in-One assistant tool for Claude Code, Codex, OpenCode, openclaw & Gemini CLI.
130.6k ★