❤️ Build. Ship. Scale. ❤️
A composable orchestration framework for agents, tools, and LLMs
⚙️ Tool orchestration meets multi-agent AI.
🧩 Modular. 🚀 Scalable. 🧠 Agentic.
CORTEX (Composable Orchestration Runtime for Tools/Agent EXchange - in LLM Applications) is an open-source framework to build:
- LLM-powered agents
- Tool servers with schema-based execution
- Agent-to-agent protocols
- Function-spec driven workflows (OpenAI, LangChain, Semantic Kernel, etc.)
| Module | Description | Status |
|---|---|---|
YAMCP |
Multi-transport plugin-based tool server |
✅ In testing / v0.1 |
A2A |
Agent-to-agent messaging protocol (queue & memory based) | 🚧 Coming Soon |
Schema |
Unified OpenAPI / Function Spec builder for LLM tool compatibility | 🧪 Drafting |
Planner |
LLM-based router that can plan multi-step calls to tools and agents | 📌 Roadmap |
- YAMCP
✅ For full documentation of the YAMCP tool server framework, see ➡️ YAMCP_README
- LLM-native architecture: schema-aware, tool-compatible, async-friendly
- Composable: Plug-and-play modules (tools, agents, queues, planners)
- Open to extension: Drop in new skills or agents with minimal boilerplate
# Clone the project
git clone https://github.com/yourname/cortex.git
cd cortex
# Go to a module (e.g., yamcp)
cd yamcp
python3 -m yamcp.main
| Feature | Description |
|---|---|
| Agent routing | Use LLMs to route user intent to correct tools/agents |
| A2A protocol | Let agents communicate, delegate, sync with memory queues |
| Agent memory | Token-level, document-level, or event memory models |
| Schema UI | Auto-gen OpenAI / LangChain tool schemas with docs |
We’re actively shaping the foundation of AI-native orchestration. We welcome:
🚀 New modules or adapters (like A2A, queue runners)
🛠 New tools via YAMCP plugin system
📢 Feature ideas, bug reports, and LLM orchestration proposals
Open a GitHub issue or start a discussion.
Created by : Biswajit Tripathy
https://github.com/BiswajitBiswa
