Awesome Agent Frameworks

Awesome Agent Frameworks

**English** | [中文](README_CN.md) | [日本語](README_JA.md) | [한국어](README_KO.md)

An architecture-focused guide to the open-source personal AI agent ecosystem (the "Claw" family).

Not a link dump. Each entry explains **what problem it solves, how it solves it, and why that approach matters**.

---

Table of Contents

• How to Read This Guide
• The Origin Story
• Architecture Map
• By Category
  • Full Platform
  • Lightweight
  • Rust Native
  • Embedded / Edge
  • Unique Paradigm
  • Python Ecosystem
  • Mobile
  • Other Languages
• Cross-Cutting Comparisons
  • Security Models
  • Channel Coverage
  • LLM Provider Strategy
  • Memory & Learning
  • Deployment Targets
• By Popularity
• Decision Framework
• Contributing

---

Disclaimer

This guide describes **architecture and design intent**, not guaranteed working software.

**Verification scope**: We cloned all repos and read source code to verify repo existence, license files, primary language, and structural claims (file counts, module organization, trait definitions). We did **not** build binaries, run tests, or verify runtime behavior. Performance numbers (binary size, RAM, boot time) are taken from project READMEs — not independently measured. Channel and tool counts are based on code definitions, not confirmed integrations.

**LOC and numeric claims**: Many projects market themselves with selective metrics. "~4K LOC" may refer to a core subset while the full codebase is 10-50x larger. "Zero unsafe code" may mean workspace-level deny with per-crate exceptions.