#4747th of 82 in Agent frameworks and SDKs
atomic-agents
Python framework for building agents from typed Pydantic input and output schemas
- Stars
- 6.3k
- License
- MIT
- Last commit
- Oct 2026
- Last release
- Oct 2026
- Language
- Python
5,725 on 4 Apr 20266,315 on 31 Oct 2026
From GitHub's daily star counts.
Checked on GitHub.
Overview
Atomic Agents is a Python library for composing LLM agents and tools from small components, each with Pydantic input and output schemas, built on Instructor. Agents combine a system prompt, history and runtime context providers, and chain by matching one component's output schema to another's input schema. A CLI, Atomic Assembler, downloads tools from the Atomic Forge collection.
Who it is for: Python developers who want typed, predictable LLM agent pipelines
Strengths
- Schema-typed inputs and outputs make agents and tools swappable and chainable
- Works with providers Instructor supports, including OpenAI, Anthropic, Gemini, Groq and Ollama
- Every agent and tool run is an OpenTelemetry span, usable with Phoenix, Langfuse or OTLP
- Atomic Forge provides ready tools such as SearXNG, Tavily, PDF Reader and Calculator
Weaknesses
- Code-only: no visual builder or hosted runtime
- Version 2.0 has breaking changes for v1.x users
- CLI downloads only tools; agents and pipelines are listed as coming later
- Non-OpenAI providers need separate Instructor extras installed
What it needs
- no GPU
- Needs Instructor, Pydantic, LLM provider API or local model server
- Models: OpenAI, Anthropic, Gemini, Groq, Ollama
Quick answers
Is atomic-agents open source?
Yes. Its code is on GitHub under the MIT license.
Does atomic-agents need a GPU?
No, it runs without one.
Also in Agent frameworks and SDKs
See all 82| Rank | Project | Score |
|---|---|---|
| 1 | ppt-masterAgent skill that turns documents into natively editable PowerPoint decks | 95 out of 100 |
| 2 | openai-agents-pythonPython SDK for multi-agent workflows with handoffs, guardrails and tracing | 90 out of 100 |
| 3 | langchainPython framework for building LLM agents and apps from interoperable components | 89 out of 100 |
| #4 | browser-usePython library that lets an LLM agent control a web browser | 89 out of 100 |
| #5 | tradingagentsMulti-agent LLM framework that simulates a trading firm's analysis process | 89 out of 100 |