agentsmesh を Homebrew Cask でインストール

プロジェクトリンク

• ホームページ
• リポジトリ
• パッケージマネージャページ

コマンドとエイリアス

ローカルパッケージデータベースに実行可能エイリアスは見つかりませんでした。

プロジェクトの歴史

AgentsMesh was created as a control layer for operating fleets of AI coding agents. Its architecture separates orchestration and monitoring from the runners that execute agents on users' machines.

The platform expanded the basic remote-agent model with AgentPods, isolated workspaces, collaboration meshes and channels, tickets, automated loops, and human takeover.

採用の歴史

AgentsMesh supports Claude Code, Codex CLI, Gemini CLI, Aider, OpenCode, and custom terminal-based agents. It offers a hosted control plane while allowing self-hosted runners so execution and source code remain on user-controlled infrastructure.

The project is distributed across web, desktop, and iOS clients, with runner packages for multiple operating systems and container images for self-hosted deployment.

使われ方

Operators install and authenticate a runner on each execution machine, then create isolated AgentPods through the web, desktop, or mobile console. Pods can be grouped for collaboration and associated with tickets and repositories.

Self-hosters can run the control-plane stack using the repository's deployment tooling. Teams bring their own model-provider API keys and may define custom agents with AgentFile.

パッケージ好きにとっての重要性

AgentsMesh treats interactive coding agents as schedulable infrastructure: each agent receives an isolated Git worktree, branch, terminal, and credential context. This addresses process isolation and state collision when many autonomous tools operate on repositories concurrently.

Its split control and data planes, self-hosted runners, AgentFile definitions, and support for multiple agent vendors make it relevant to developers building reproducible multi-agent workflows rather than adopting a single assistant.

タイムライン

• 2026: Official documentation described a hosted and self-hosted workforce platform spanning web, desktop, and mobile clients.
• 2026: The platform documented AgentPods, runner fleets, Git-worktree isolation, collaboration meshes, tickets, and automated control loops.

Related projects

• Built-in integrations include Claude Code, Codex CLI, Gemini CLI, Aider, and OpenCode.
• AgentFile is the project's declarative format for specifying an agent's command, environment, configuration, and build behavior.
• GitHub, GitLab, and Gitee integrations connect repositories and development workflows to agent workspaces.

ソース

• Official API documentation: https://github.com/AgentsMesh/AgentsMesh/blob/main/docs/api/README.md
• Official AgentFile reference: https://agentsmesh.ai/docs/concepts/agentfile
• Official documentation: https://agentsmesh.ai/docs
• Official repository and README: https://github.com/AgentsMesh/AgentsMesh