Skip to main content

​
What is Splinter?

Splinter is control and coordination infrastructure for multi-agent AI systems. It sits between your agents and LLM providers, ensuring your agents don’t go rogue.

Control Layer

Runtime controls. Budget limits, rate limiting, circuit breakers, decision locks.

Coordination Layer

Multi-agent collaboration. Shared state, checkpointing, handoffs, goal tracking.

​
The Problem

Without proper controls, AI agents will:
  • Spend unlimited money — A single agent can rack up thousands in API costs
  • Loop forever — Agents get stuck repeating the same actions
  • Overwrite each other — Multiple agents writing to the same state = chaos
  • Crash with no recovery — Hours of work lost when something fails

​
The Solution

from splinter import Splinter

s = Splinter(
    openai_key="sk-...",
    max_budget=5.0,   # Stop at $5
    max_steps=50,     # Stop after 50 calls
)

result = await s.run("researcher", "Research AI trends")
print(f"Cost: ${s.cost:.4f}")  # Always know what you spent
Splinter prevents all of these problems with a simple, pluggable architecture.

​
Architecture

​
Local vs Cloud

Local (Free)

All Control and Coordination features run entirely on your machine. No external dependencies.

Cloud (Paid)

Add an API key for live dashboard, remote control, deadlock detection, and more.

Quickstart

Get up and running in 5 minutes

Concepts

Understand the core ideas

API Reference

Full API documentation