Agensh

Agensh is a decentralized multi-agent system that scales up to 1,024 concurrent workers for collaborative tasks.

Agensh is a decentralized multi-agent system that scales to 1,024 agents to handle complex tasks asynchronously. Agents discover and claim sub-tasks, share findings, verify results, and merge progress without a central orchestrator. A shared workspace, message interface, and shared context connect each agent's local progress to the team. The system integrates Gitea for the shared workspace, Mattermost for messaging, and utilizes context adapted from DeLM. It operates with GPT models including GPT-5.6-sol and GPT-6.1-sol using Copilot ACP sessions and Docker Compose.

Agensh at a glance

Best for: Software engineer
Runs on: CLI, self-hosted, open source
Pricing: Free
Released: October 2026
Category: AI agent tool
Website: github.com/microsoft/agensh

Key features

Scales up to 1,024 agents
Decentralized peer coordination
Asynchronous progress merging
Shared workspace via Gitea
Messaging interface via Mattermost
Shared context mechanism

Use it for

Rebuilding reference programs from behavioral observations
Running multi-agent software engineering experiments
Evaluating test-pass rates across large agent teams

Alternatives to Agensh

Memoh — Agents run 24/7 on dedicated cloud computers with desktops and persistent storage.
Agent Executor — Declarative runtime for scaling stateful agent workloads with isolation.
AgentENV — Scales massive Firecracker environments across machines efficiently.
Alook — Connects existing local coding agents to team communication channels.
AgentDock — It coordinates multi-device workflows and provides secure environment access.
Swarms — Provides production-ready infrastructure with a 99.9% uptime guarantee.
Dsh Agent Teams — Assembles sub-agents
Agor — Enables teams to collaborate on AI agent work using live spatial boards and shared git branches.
Starnet — Visual agent management with transparent, persistent on-disk logs and state.
n8n — Self-hostable automation with code allowed in any step
Raven — Provides a persistent, self-evolving environment for all-domain agent collaboration.
Yao — Self-hosted agent platform with task tracking on centralized boards.

All Agensh alternatives