Build AI agents
that just work.
A TypeScript framework for reliable AI agents with full transparency and control. Inspired by the Finnish concept of sisu — calm determination under pressure.
No surprises
Explicit middleware, typed tools, deterministic control flow. Exactly what you expect, every time.
Full control
Compose planning, routing, and safety like Express apps. Your pipeline, your rules.
Total visibility
Built-in tracing, logging, and debugging out of the box. See every token, cost, and decision.
Provider-agnostic
OpenAI, Anthropic, Ollama, or bring your own. Swap providers in a single line of code.
Stop fighting your
AI framework.
Sisu was built because debugging AI agents shouldn't require a PhD. Every decision is visible, every tool is typed, every failure is traceable.
Up and running in minutes.
Install, wire up your first agent, and have full tracing before your first coffee is finished.
1 — Install
pnpm add @sisu-ai/core @sisu-ai/adapter-openai \
@sisu-ai/mw-register-tools @sisu-ai/mw-tool-calling \
@sisu-ai/mw-conversation-buffer @sisu-ai/mw-trace-viewer \
@sisu-ai/mw-error-boundary zod dotenv3 — Run
npx tsx agent.ts
open traces/viewer.html2 — Your first agent
import "dotenv/config";
import { Agent, createCtx } from "@sisu-ai/core";
import { registerTools } from "@sisu-ai/mw-register-tools";
import { conversationBuffer } from "@sisu-ai/mw-conversation-buffer";
import { errorBoundary } from "@sisu-ai/mw-error-boundary";
import { toolCalling } from "@sisu-ai/mw-tool-calling";
import { openAIAdapter } from "@sisu-ai/adapter-openai";
import { traceViewer } from "@sisu-ai/mw-trace-viewer";
import { z } from "zod";
const ctx = createCtx({
model: openAIAdapter({ model: "gpt-4o-mini" }),
input: "What is the weather in Stockholm?",
systemPrompt: "You are a helpful assistant.",
});
const app = new Agent()
.use(errorBoundary())
.use(traceViewer())
.use(registerTools([weather]))
.use(conversationBuffer({ window: 8 }))
.use(toolCalling);
await app.handler()(ctx);Two powerful tools.
One coherent system.
SISU Framework
@sisu-ai/coreA TypeScript framework that brings Express-style middleware composition to AI agent development. Typed tools, deterministic control flow, built-in tracing, and provider-agnostic adapters for OpenAI, Anthropic, and Ollama.
- ✓ Middleware pipeline (errorBoundary, toolCalling, guardrails…)
- ✓ HTML & CLI trace viewer
- ✓ Control flow: sequence, branch, parallel, graph
- ✓ RAG, RAG tools, vector adapters
- ✓ 25+ runnable examples
SISU CLI
@sisu-ai/cliA configurable agent CLI for real terminal work. Run interactive chat, tune providers and tool policy, install skills, and quickly discover Sisu framework packages when you need building blocks.
- ✓ Interactive
sisu chatfor everyday terminal workflows - ✓ Scaffold: chat-agent, cli-agent, rag-agent
- ✓ Package discovery: list, info, install
- ✓ Tool safety model with policy gating
- ✓ Session persistence and branching
Everything you need, nothing you don't.
Modular packages. Use only what your agent needs.