Quick Start

Get AgentsMesh up and running in six steps. This is a quick overview — each step links to a detailed tutorial with troubleshooting and advanced options.

1

Create an Account

Register at the sign-up page with your email, username, and password. After submitting you will receive a verification email — check your spam folder if it does not arrive within a few minutes.

After verifying your email, the onboarding wizard will guide you to:

  • Create a personal workspace (one click) or a team workspace
  • Optionally join an existing organization via an invitation link
2

Setup a Runner

A Runner is a lightweight daemon that runs on your machine and executes AI agents in isolated workspaces called Pods. Install the AgentsMesh Runner on your development machine or server.

# Download and install the runner
curl -fsSL https://agentsmesh.ai/install.sh | sh

Register with a token

agentsmesh-runner register \
  --server https://agentsmesh.ai \
  --token <YOUR_TOKEN>
agentsmesh-runner run

Register with browser login

agentsmesh-runner login
agentsmesh-runner run

See Runner Setup for Docker, Linux packages, and Kubernetes deployment options.

3

Install AI Agent CLIs

Install the CLI tools you want to use on your Runner machine and set the required API key environment variables. The Runner auto-detects installed CLIs via PATH and reports available agents to the platform.

Claude Code

npm install -g @anthropic-ai/claude-code
export ANTHROPIC_API_KEY=sk-ant-...

Get your API key from console.anthropic.com

Codex CLI

npm install -g @openai/codex
export OPENAI_API_KEY=sk-...

Get your API key from platform.openai.com

Gemini CLI

npm install -g @google/gemini-cli
export GOOGLE_API_KEY=...

Get your API key from aistudio.google.com

After installing, restart the Runner or wait for the next heartbeat. The newly available agents will appear in the New Pod dialog.

See MCP & Skills for more details on MCP tools and skill configuration.

4

Connect a Git Provider

Connecting a Git provider lets your Pods clone repositories and work in isolated Git worktrees. Go to Settings → Personal → Git Settings to add a provider.

GitHub

Add a Personal Access Token with the repo scope. github.com/settings/tokens

GitLab

Add a Personal Access Token with api and read_repository scopes. Self-hosted instances are supported.

See Connect Git Repos for full setup instructions including SSH keys and Gitee.

5

Start a Pod

A Pod (AgentPod) is a virtual workstation where an AI agent runs with its own terminal and isolated code workspace. Create your first one now!

  1. Navigate to Workspace in the sidebar
  2. Click New Pod
  3. Select an AI agent (Claude Code, Codex CLI, etc.)
  4. Choose a Runner
  5. Optionally select a repository and branch — the agent will work in an isolated copy (Git worktree) so your main branch stays safe
  6. Optionally link to a ticket for context
  7. Add an initial prompt describing the task (e.g., "Add input validation to the login form" or "Fix the failing test in auth.test.ts")
  8. Click Create

See Your First Pod for a detailed walkthrough with troubleshooting.

6

Interact with Your Agent

After creation the terminal view opens with a live session. Your AI agent begins executing the initial prompt or waits for input if none was provided.

  • Type follow-up prompts or commands directly in the terminal
  • Watch the agent status indicator — Idle (ready for input), Executing (working on your task), or Waiting (needs your approval)
  • The agent works inside the repository worktree if one was linked
  • Terminate the Pod from the toolbar when your task is complete

Try It: Your First Pod in 60 Seconds

Want to see it work right away? Here's the fastest path — no repository needed:

  1. Click Workspace → New Pod
  2. Select Claude Code (or any installed agent) and your Runner
  3. Skip the repository and ticket fields
  4. Set the initial prompt to: "Say hello, then list the files in the current directory and explain what you see."
  5. Click Create and watch the agent respond in the terminal
This creates a Pod without a repository — the agent runs in a temporary directory. It's perfect for testing your setup before working with real code.

Next Steps