ClawGo Docs

English

简体中文

English

简体中文

Appearance

ClawGoA Production-Oriented, Go-Native Agent Runtime

Long-running, observable, recoverable, orchestrated runtime for real agent systems

Quick Start
Architecture
ClawGo Logo

🕸️ Multi-Agent Topology

A unified `main / subagents / remote branches` model where internal collaboration stays observable without polluting user-facing chat.

♻️ Recoverable Execution

`subagent_runs.jsonl`, `subagent_events.jsonl`, `threads.jsonl`, and `agent_messages.jsonl` allow recovery after restart.

🛠️ Operational Engineering

Use `config.json`, `AGENT.md`, logs, memory, nodes, OAuth, and WebUI as a full runtime control surface.

🦞 What ClawGo Is ​

ClawGo is currently an Agent Runtime, not a world simulator and not just a chat wrapper.

The main model is:

  • main agent handles entry, routing, dispatch, and merge
  • subagent runtime executes local and remote branches
  • runtime store persists runs, events, threads, messages, and memory
  • WebUI focuses on inspection, status views, and account management

Default collaboration flow:

text
user -> main -> worker -> main -> user

📚 What This Documentation Covers ​

  • installation, onboarding, provider selection, and provider login
  • the agents.subagents registry and tool-permission model
  • subagent run, thread, and message persistence
  • CLI, Gateway, WebUI, nodes, MCP, cron, and channels
  • local development, build, deployment, and workspace templates

🗂️ Documentation Sections ​

🧠 Concepts ​

🚀 Guides ​

📖 Reference ​

🔧 Ops & Dev ​

✅ Suggested Reading Order ​

  1. Quick Start
  2. Architecture
  3. Configuration
  4. Runtime, Storage, and Recovery
  5. Subagents and Skills

📮 Contact ​