Pi Mono: AI agent toolkit and LLM API
Monorepo providing AI agent development tools, unified LLM API, and deployment management for multiple providers.
Learn more about Pi Mono
Pi Mono is a TypeScript monorepo containing tools for building AI agents and managing LLM deployments. The toolkit includes a unified API client supporting multiple LLM providers (OpenAI, Anthropic, Google), an agent runtime with tool calling capabilities, and UI libraries for both terminal and web interfaces. It features a coding agent CLI that can interact with codebases, a Slack bot integration, and utilities for managing vLLM deployments on GPU infrastructure. The modular architecture allows developers to use individual packages or combine them for complete AI agent solutions.
Unified LLM API
Single interface for multiple LLM providers including OpenAI, Anthropic, and Google. Abstracts provider-specific implementations behind a consistent API.
Modular Architecture
Seven separate npm packages that can be used independently or together. Includes agent runtime, UI libraries, deployment tools, and integrations.
Multi-Interface Support
Provides terminal UI, web components, CLI tools, and Slack bot integration. Supports both interactive and programmatic usage patterns.
import { Agent } from '@mariozechner/pi-agent-core';
import { OpenAIProvider } from '@mariozechner/pi-ai';
const agent = new Agent({
provider: new OpenAIProvider({ apiKey: process.env.OPENAI_API_KEY }),
model: 'gpt-4',
tools: [
{
name: 'get_weather',
description: 'Get current weather for a location',
parameters: {
type: 'object',
properties: {
location: { type: 'string', description: 'City name' }
},
required: ['location']
},
handler: async ({ location }) => {
const response = await fetch(`https://api.weather.com/v1/current?location=${location}`);
return response.json();
}
}
]
});
const response = await agent.chat('What\'s the weather in San Francisco?');
console.log(response.content);v0.54.0: Added - Added default skill auto-discovery for `.agents/skills` locations
- –Added - Added default skill auto-discovery for `.agents/skills` locations
- –Pi now discovers project skills from `.agents/skills` in `cwd` and ancestor directories (up to git repo root, or filesystem root when not in a repo), and global skills from `~/.agents/skills`, in addition to existing `.pi` skill paths
v0.53.0: Added `SettingsManager.drainErrors()` for caller-controlled settings I/O error handling without mana
- –Added `SettingsManager.drainErrors()` for caller-controlled settings I/O error handling without manager-side console output.
- –Added auth storage backends (`FileAuthStorageBackend`, `InMemoryAuthStorageBackend`) and `AuthStorage.fromStorage(...)` for storage-first auth persistence wiring.
- –Added Anthropic `claude-sonnet-4-6` model fallback entry to generated model definitions.
- –`SettingsManager` now uses scoped storage abstraction with per-scope locked read/merge/write persistence for global and project settings.
- –Fixed project settings persistence to preserve unrelated external edits via merge-on-write, while still applying in-memory changes for modified keys.
v0.52.12: Added `transport` setting (`"sse"`, `"websocket"`, `"auto"`) to `/settings` and `settings.json` for
- –Added `transport` setting (`"sse"`, `"websocket"`, `"auto"`) to `/settings` and `settings.json` for providers that support multiple transports (currently `openai-codex` via OpenAI Codex Responses).
- –Interactive mode now applies transport changes immediately to the active agent session.
- –Settings migration now maps legacy `websockets: boolean` to the new `transport` setting.
See how people are using Pi Mono
Top in AI & ML
Related Repositories
Discover similar tools and frameworks used by developers
Higgsfield
Cluster manager for multi-node PyTorch model training.
KoboldCpp
Self-contained llama.cpp distribution with KoboldAI API for running LLMs on consumer hardware.
llama_index
Connect LLMs to external data via RAG workflows.
llama.cpp
Quantized LLM inference with hardware-accelerated CPU/GPU backends.
ByteTrack
Multi-object tracker associating low-confidence detections across frames.
Related Reading
Guides and comparisons from the Greptile content library
Best SAST Tools for Engineering Teams [2026 Review]
We compare five of the best SAST tools (Semgrep, Checkmarx One, Snyk, Aikido, and GitHub CodeQL) on core functionality, languages, pricing, and fit, plus why static scanning needs runtime testing and full-codebase review alongside it.
What is Source Code Management (SCM): Everything Engineering Leaders Need to Know
Source code management is more than branching and merging. Here is how SCM works, the top tools, the best practices, and why good code review is the foundation of source control in an agentic era.
27 Code Quality Tools that Catch Bugs [2026 Review]
The best code quality tools that catch bugs, security vulnerabilities, and runtime errors, sorted by language, with type, noise level, cost, and who each one is for.
Guide to Self-Hosted AI Code Review for GitLab and GitHub Enterprise [2026]
How self-hosted AI code review works on GitLab Self-Managed and GitHub Enterprise Server, how it compares to SaaS review tools, what to look for in a self-hosted tool, and what a self-hosted rollout actually takes.