Nous
open sourceTypeScript agent framework with autonomous coding agents, WebUI, and LLM-independent function calling
Updated 2026-03-17
Nous is an open-source agent framework built for TypeScript developers that combines multi-agent orchestration, autonomous coding agents, and observability into a single integrated platform. It features LLM-independent function calling with auto-generated schemas, a WebAssembly-sandboxed Python execution layer via pyodide, database persistence, tracing, and a web UI, designed to reduce token costs and latency in frontier LLM interactions.
Use Cases
- ·Building autonomous DevOps/SRE agents for infrastructure automation
- ·Creating autonomous coding agents that reason across multiple LLM calls
- ·Implementing multi-agent workflows with persistent state and observability
- ·Reducing LLM API costs by batching function calls and validation in a single control loop
- ·Building solo AI tools for GitLab/GitHub automation and code review
Key Points
- ·TypeScript-native framework combining CrewAI, OpenDevin, and LangFuse-like features in one platform
- ·Autonomous agent generates and executes Python code in WebAssembly sandbox (pyodide) for safe multi-step reasoning
- ·LLM-agnostic function calling with auto-generated schemas reduces vendor lock-in
- ·Built-in observability: database persistence, tracing, Web UI, and human-in-the-loop validation
- ·Designed to optimize frontier LLM usage by batching multiple calls and validation logic per control loop
- ·Includes practical DevOps tooling: configurable GitLab merge request AI reviewer
- ·Lower barrier to entry than LangChain with batteries-included architecture
Tags
typescriptautonomous-agentscoding-agentsfunction-callingwebassemblyopen-sourcedevops-automation
Do-Nothing Score
Find out how close you are to Ghost CEO.
Machine-Readable
This tool is available as structured JSON via the API.
GET /api/v1/tools/nous-open-source-agent-framework