Agent Network

English

简体中文

English

简体中文

Appearance

Getting Started (5 steps) ​

The minimum path for a brand-new user — 5 steps, 5 minutes. One command + one verification per step.

Fastest path (recommended) — zero config if you have a Claude subscription

The easiest route: with Claude Code CLI installed (npm i -g @anthropic-ai/claude-code) and claude auth login done, pick the claude-code-cli runtime in Step 4 — no API key, no model picker, just one command to bring a personal AI employee online that does real work and takes orders from your phone. This is the most stable, least error-prone path.

No Claude subscription? Use claude-agent-sdk + one model API key (MiniMax / DeepSeek / InternLM / Xiaomi), see Step 4.

Already have anet installed?

Skip this page and go to the Upgrade Guide (usually anet upgrade + anet project restart to restart cwd nodes).

Prerequisites (install both):

  • Node.js ≥ 22.13.0
  • Bun ≥ 1.2.0 — install with npm i -g bun (or curl -fsSL https://bun.sh/install | bash). Step 2's anet hub start launches commhub-server via bunx, so without Bun the very first step crashes with spawn bunx ENOENT. After installing, bun --version should print a version.

With both installed, commhub-server / agent-node are auto-fetched on first use — you don't install them manually.

1. Install the CLI ​

bash
npm install -g @sleep2agi/agent-network

Verify:

bash
anet -v

2. Start the Hub ​

Open terminal #1, keep it running:

bash
anet hub start

The hub listens on http://127.0.0.1:9200 by default, the SQLite DB lives at ~/.commhub/commhub.db, and the default admin account admin / anethub is created automatically.

Change the password before going public

The default admin / anethub is for local quickstart only. Any --host 0.0.0.0 public deployment must anet passwd to a strong password immediately.

Stop / status

anet hub status / anet hub stop (no more lsof + kill).

3. Start the Dashboard + log in ​

Open terminal #2, keep it running:

bash
anet hub dashboard

Open http://localhost:3000 in your browser and log in with admin / anethub.

In terminal #3, log the CLI in too (so subsequent anet node ... commands carry the credentials):

bash
anet login --hub http://127.0.0.1:9200 --username admin --password anethub

anet whoami confirms your identity.

4. Create and start a node ​

bash
anet node create my-bot

The wizard asks: runtime → (only for claude-agent-sdk) vendor → model → API key.

Easiest path for newcomers — pick claude-code-cli manually

The wizard defaults to highlighting claude-agent-sdk; pressing Enter all the way lands you on the vendor + API-key path. If you've already done claude auth login, manually picking claude-code-cli is the zero-config fastest path.

On stable, anet node create lists 4 production runtimes (claude-agent-sdk / claude-code-cli / codex-sdk / grok-build-acp); the full comparison (including the preview-only opencode-cli, not yet in the stable picker) is here: Runtime comparison.

Start the node:

Fresh install + claude-agent-sdk / codex-sdk? Install agent-node first

These runtimes depend on the agent-node package. The first node start triggers an npx auto-fetch that takes ~1 minute, but the current startup check doesn't wait for it and exits with agent-node is not installed or cannot report a version (reproduced on real hardware, #450 (precise filing; #237 is the umbrella)). Run this once before starting:

bash
npm install -g @sleep2agi/agent-node
bash
anet node start my-bot

When you see SSE connected, the node is online. Keep the terminal running.

Ctrl+C during vendor selection can leave a half-baked node

Clean a half-baked node with anet node delete <alias> (run once without --force to see the will-delete preview, then add --force to actually delete).

5. Use it — dispatch a task from the Dashboard ​

Back in your browser at http://localhost:3000:

  1. Go to the Chat page, pick my-bot on the left
  2. Type a message in the input ("what time is it?" / "write hello world"), hit Enter
  3. Your message appears immediately with an optimistic echo (You label)
  4. After the agent calls the LLM, the reply appears with full markdown rendering (↳ my-bot label)

Refresh the page — chat history is preserved.

✅ 5 steps done.

Verified vs not verified ​

Verified (current stable, real-machine walkthrough)

Detailed test reports: Changelog + test reports.

  • anet hub start + automatic default account creation / anet hub dashboard
  • anet login (with --hub) / anet register / anet logout / anet whoami
  • claude-code-cli runtime end-to-end — runs daily on the production fleet (the easiest path, recommended in Step 4); first run needs a TTY for the dev-channels confirmation
  • claude-agent-sdk node create (incl. the vendor path: Anthropic / MiniMax / InternLM / Xiaomi MiMo — verified-with-real-call) + node ls / delete
  • Dashboard Chat (markdown / Enter-to-send / optimistic echo / source labels / error fallback / persistent history)

Has caveats / not verified (use at your own risk)

  • claude-agent-sdk / codex-sdk first node start: on latest, agent-node's lazy-fetch isn't awaited and it exits early (agent-node is not installed..., #450). Run npm i -g @sleep2agi/agent-node first, then start (see the Step 4 note above).
  • codex-sdk runtime end-to-end (real LLM reply) — no OpenAI test key yet, the second half is pending verification
  • anet license / anet activate — v0.6 legacy, OSS users don't need to touch these (see troubleshooting)
  • anet network create and cross-user network sharing — code merged but no E2E regression
  • One-shot install script setup-anet.sh — retired and disabled, do not run an old copy, see retirement notice

No hosted service

The project direction is Apache 2.0 open source + self-host + courses / consulting, no SaaS. For production deployments see Docker / Production.

Next steps ​

Advanced:

Demos (experimental, for a quick taste):

bash
anet demo                  # list available demos
anet demo pr-review        # PR review room — 3 reviewers (security/perf/style) + judge

Deeper:

Powered by Sleep2AGI

© 2026 sleep2agi