The AI agent stack
Autonomous agents plan, call tools, and act over multiple steps. The stack pairs an agent runtime with model tooling, integration glue, and tracing.
Agent runtime - planning, tool use, and multi-step control flow.
When not to use: Don't use an agent loop when a deterministic workflow would do; agents add latency, cost, and non-determinism.
ECC
affaan-m/ECC
The agent harness performance optimization system. Skills, instincts, memory, security, and research-first development for Claude Code, Code
hermes-agent
NousResearch/hermes-agent
The agent that grows with you
AutoGPT
Significant-Gravitas/AutoGPT
AutoGPT is the vision of accessible AI for everyone, to use and to build on. Our mission is to provide the tools, so that you can focus on w
ollama
ollama/ollama
Get up and running with Kimi-K2.6, GLM-5.1, MiniMax, DeepSeek, gpt-oss, Qwen, Gemma and other models.
Framework & tooling - prompt/state management and RAG the agent can call.
When not to use: Skip the framework layer if your agent only needs one or two hardcoded tools.
ECC
affaan-m/ECC
The agent harness performance optimization system. Skills, instincts, memory, security, and research-first development for Claude Code, Code
hermes-agent
NousResearch/hermes-agent
The agent that grows with you
AutoGPT
Significant-Gravitas/AutoGPT
AutoGPT is the vision of accessible AI for everyone, to use and to build on. Our mission is to provide the tools, so that you can focus on w
ollama
ollama/ollama
Get up and running with Kimi-K2.6, GLM-5.1, MiniMax, DeepSeek, gpt-oss, Qwen, Gemma and other models.
Gateways & SDKs - route across model providers and expose tools (e.g. via MCP).
When not to use: A gateway is overkill when you're pinned to a single provider and model.
JavaGuide
Snailclimb/JavaGuide
Java 面试 & 后端通用面试指南，覆盖计算机基础、数据库、分布式、高并发、系统设计与 AI 应用开发
dify
langgenius/dify
Production-ready platform for agentic workflow development.
firecrawl
firecrawl/firecrawl
The API to search, scrape, and interact with the web at scale. 🔥
browser-use
browser-use/browser-use
🌐 Make websites accessible for AI agents. Automate tasks online with ease.
Tracing - see every step, tool call, and token so failures are debuggable.
When not to use: There is no good reason to run agents in production without tracing.
toon
toon-format/toon
🎒 Token-Oriented Object Notation (TOON) – Compact, human-readable, schema-aware JSON for LLM prompts. Spec, benchmarks, TypeScript SDK.
VAR
FoundationVision/VAR
[NeurIPS 2024 Best Paper Award][GPT beats diffusion🔥] [scaling laws in visual generation📈] Official impl. of "Visual Autoregressive Modeli
LLM4Decompile
albertan017/LLM4Decompile
Reverse Engineering: Decompiling Binary Code with Large Language Models
tree-of-thought-llm
princeton-nlp/tree-of-thought-llm
[NeurIPS 2023] Tree of Thoughts: Deliberate Problem Solving with Large Language Models