Back to search
Anemoi
SkillGitHub
Anemoi: A Semi-Centralized Multi-agent Systems Based on Agent-to-Agent Communication MCP server from Coral Protocol
Safety scan
A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.
Checked today
Badge for your README
Markdown
[](/listings/coral-protocol--anemoi)HTML
<a href="/listings/coral-protocol--anemoi"><img src="/api/listings/coral-protocol--anemoi/badge.svg" alt="findsafeskills safety rating" /></a>Install
i
What was scanned
The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.
README.md · README · 96 lines
View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)
- # Anemoi: A Semi-Centralized Multi-agent Systems Based on Agent-to-Agent Communication MCP server from Coral Protocol
- **Anemoi** is a semi-centralized multi-agent system (MAS) built on a **Agent-to-Agent (A2A) communication MCP server**.
- Unlike traditional **context-engineering + centralized** paradigms, Anemoi introduces **structured, direct inter-agent communication** — enabling agents to collaborate much like a real-world team.
- 🌀 *Like winds connecting distant lands, Anemoi enables agents to communicate directly in a semi-centralized network, achieving scalable coordination and seamless information flow.*
- <p align="center">
- <img src="Anemoi/images/Anemoi_semi.png" alt="Anemoi Concept" width="70%">
- </p>
- ---
- ## 🚀 Key Features
- * **Semi-Centralized Architecture**
- Reduces dependency on a single planner agent, supporting adaptive plan updates.
- * **Direct Agent-to-Agent Collaboration**
- Agents can monitor progress, assess results, identify bottlenecks, and propose refinements in real time.
- * **Efficient Context Management**
- Minimizes redundant prompt concatenation and information loss, improving scalability and cost-efficiency.
- * **Benchmark Performance**
- Achieved **52.73% accuracy** on the validation set of the GAIA benchmark, setting the **state-of-the-art** among small-LLM-based systems.
- > Surpasses OWL (43.63%) by **+9.09%** in the same worker agents and models configuration (gpt-4.1-mini as planner agent/ gpt-4o as worker agent).
- <p align="center">
- <img src="Anemoi/images/Anemoi_workflow.png" alt="Anemoi Workflow" width="85%">
- </p>
- ---
- ## 📄 Publication
- Our work has been released on arXiv:
- 👉 [Anemoi: A Semi-Centralized Multi-agent Systems Based on Agent-to-Agent Communication MCP server from Coral Protocol](https://arxiv.org/abs/2508.17068)
- If you find this project useful, please consider citing our paper:
- ```
- @article{ren2025anemoi,
- title={Anemoi: A Semi-Centralized Multi-agent Systems Based on Agent-to-Agent Communication MCP server from Coral Protocol},
- author={Ren, Xinxing and Forder, Caelum and Zang, Qianbo and Tahir, Ahsen and Georgio, Roman J. and Deb, Suman and Carroll, Peter and Gürcan, Önder and Guo, Zekun},
- journal={arXiv preprint arXiv:2508.17068},
- year={2025},
- url={https://arxiv.org/abs/2508.17068}
- }
- ```
- ---
- ## 🧪 Reproduction
- Set up environment variables:
- ```
- echo '
- export FIRECRAWL_API_KEY="your_firecrawl_api_key"
- export GOOGLE_API_KEY="your_google_api_key"
- export HF_HOME="your_hf_home_path"
- export OPENROUTER_API_KEY="your_openrouter_api_key"
- export SEARCH_ENGINE_ID="your_search_engine_id"
- export CHUNKR_API_KEY="your_chunkr_api_key"
- ' >> ~/.bashrc && source ~/.bashrc
- ```
- Create environment:
- ```
- cd Anemoi
- /usr/bin/python3.12 -m venv venv
- source venv/bin/activate
- pip install -r requirements.txt
- ```
- We made some minor modifications to CAMEL 0.2.70 for our experiments:
- ```
- rm -rf venv/lib/python3.12/site-packages/camel
- cp -r utils/camel venv/lib/python3.12/site-packages/
- ```
- Run the experiment:
- ```
- cd ..
- ./gradlew run --console=plain
- ```
Ratings and comments
Sign in to rate this skill.
Sign inFree to use. If it’s helped, pass it on.