Links
Maintainers
- DEEP-IOS (https://github.com/DEEP-IOS)
Purpose
Claw-Swarm is a swarm intelligence plugin that enables multiple AI agents to collaborate like a bee colony. It uses 12 bio-inspired algorithms across a 6-layer architecture to solve real problems in multi-agent coordination:
What it does:
- ๐ Pheromone-based indirect communication โ Agents leave "scent trails" (MMAS ant colony algorithm) so others can discover relevant context without explicit messaging
- ๐ง Intelligent memory management โ Three-tier working memory + Ebbinghaus forgetting curve for automatic pruning of unimportant memories
- ๐ฏ Self-organizing task allocation โ FIPA Contract Net Protocol for negotiation + ABC (Artificial Bee Colony) scheduling for explore/exploit balance
- ๐ Automatic role discovery โ k-means++ clustering on agent behavior to discover emergent roles, plus MoE (Mixture of Experts) routing for role-task matching
- ๐ Real-time monitoring dashboard โ Dark-theme UI with RED metrics, agent states, and pheromone visualization via SSE
The 12 algorithms: MMAS, ACO Roulette, Ebbinghaus Forgetting Curve, BFS Knowledge Graph, PARL A/B Persona Evolution, GEP Execution Planning, CPM Critical Path, Jaccard Similarity, MoE Expert Routing, FIPA Contract Net, ABC Bee Colony Scheduling, k-means++ Role Discovery
Quick start:
npm install openclaw-swarm
cd node_modules/openclaw-swarm
node install.js
openclaw gateway restart
Stats: 55+ source files | 34 DB tables | 490 tests (< 2s) | 382 KB package | Node.js >= 22 required
Happy to answer any questions! ๐