#5050th of 54 in Agent frameworks and SDKs
pocketflow
Python LLM framework built on a 100-line graph abstraction
- Stars
- 11.3k
- License
- MIT
- Last commit
- Jul 2026
- Language
- Python
10,155 on 31 Mar 202611,269 on 20 Oct 2026
From GitHub's daily star counts.
Checked on GitHub.
Overview
Pocket Flow is a Python library whose core is a single 100-line file implementing a graph of nodes and flows. Agents, workflows, RAG, batch and map-reduce patterns are built on top of it. It ships no vendor or app-specific wrappers, so LLM calls and tools are written by the user, and a large cookbook shows the patterns.
Who it is for: Python developers who want a minimal base for custom LLM agents and workflows
Strengths
- Core is one 100-line file with zero dependencies; can be copied into a project
- No vendor-specific wrappers, so no lock-in to a provider or vector store
- Cookbook with about 40 runnable examples: agents, RAG, MCP, A2A, voice, parallel batch
- Ports exist for TypeScript, Java, C++, Go, Rust and PHP
Weaknesses
- No built-in LLM, vector DB or tool integrations; you write each utility function
- Size and line-count comparison table is the project's own, not independently verified
- No tagged releases listed; installed from PyPI or by copying source
- Minimal abstraction means no built-in tracing, UI or deployment tooling
What it needs
- no GPU
- Models: any (user-written LLM calls)
Also in Agent frameworks and SDKs
See all 54| Rank | Project | Score |
|---|---|---|
| 1 | ppt-masterAgent skill that turns documents into natively editable PowerPoint decks | 94 out of 100 |
| 2 | langchainPython framework for building LLM agents and apps from interoperable components | 89 out of 100 |
| 3 | browser-usePython library that lets an LLM agent control a web browser | 89 out of 100 |
| #4 | tradingagentsLangGraph multi-agent framework simulating a trading firm to analyze tickers | 88 out of 100 |
| #5 | copilotkitSDK for adding agent chat, generative UI and shared state to apps | 87 out of 100 |