ClawGo: High-Performance AI Agent (Linux Server Only)

中文

ClawGo is a high-performance AI assistant tailored for Linux servers. Leveraging the concurrency advantages and binary distribution of the Go language, it provides full agent capabilities with minimal resource overhead.

🚀 Key Advantages

• ⚡ Native Performance: Optimized specifically for Linux server environments, with no dependency on Node.js or Python.
• 🏗️ Production-Ready: Single binary deployment, perfectly integrates with service management tools like systemd.
• 🔌 Mandatory Upstream Proxy: Centralized management of model quotas and authentication via CLIProxyAPI.
• 🧩 Powerful Skill Extension: Built-in productivity tools like coding-agent, github, and context7.

🏁 Quick Start

1. Initialize

clawgo onboard

When running clawgo gateway, a yes/no prompt asks whether to grant root privileges. If yes, the command is re-executed via sudo and a high-permission shell policy is enabled (with rm -rf / still hard-blocked).

2. Configure CLIProxyAPI ClawGo requires CLIProxyAPI as the model access layer.

clawgo login

3. Start Running

# Interactive mode
clawgo agent

# Gateway mode (supports Telegram, Discord, etc.)
clawgo gateway

# Gateway service management
clawgo gateway start
clawgo gateway restart
clawgo gateway stop

⚙️ Config Management & Hot Reload

ClawGo can update config.json from CLI and trigger hot reload for a running gateway:

# Set config value (supports enable -> enabled alias)
clawgo config set channels.telegram.enable true

# Read config value
clawgo config get channels.telegram.enabled

# Validate config
clawgo config check

# Trigger hot reload manually (sends SIGHUP to gateway)
clawgo config reload

Global custom config path:

clawgo --config /path/to/config.json status

Or via environment variable:

export CLAWGO_CONFIG=/path/to/config.json

config set now uses atomic write, and if gateway is running but hot reload fails, it rolls back to backup automatically.

Slash commands are also supported in chat channels:

/help
/stop
/status
/config get channels.telegram.enabled
/config set channels.telegram.enabled true
/reload

Autonomy control commands (optional; natural language is recommended):

/autonomy start [idle]
/autonomy stop
/autonomy status
/autolearn start [interval]
/autolearn stop
/autolearn status

Message scheduling policy (per session_key):

• Same session runs in strict FIFO order; later messages are queued.
• /stop immediately cancels the current response, then processing continues with the next queued message.
• Different sessions can run concurrently.

🧭 Autonomy Mode & Natural-Language Control

• Autonomy/auto-learning controls are interpreted with LLM semantic parsing first (multi-language), instead of fixed keyword matching.
• Rule-based parsing is used only as a fallback (for explicit forms such as /autonomy ... and /autolearn ...).
• If a focus direction is provided when starting autonomy mode, execution prioritizes that focus; when user says it is complete, the agent switches to other high-value tasks automatically.
• Progress updates are natural-language messages, not rigid stage-number templates.

At startup, AGENTS.md, SOUL.md, and USER.md are loaded as behavior constraints and semantic context.

🧩 Onboard/Install Doc Sync

• Both clawgo onboard and make install sync AGENTS.md, SOUL.md, USER.md into workspace.
• If a file does not exist: it is created.
• If a file already exists: only the CLAWGO MANAGED BLOCK section is updated; user custom content is preserved (incremental update, no whole-file overwrite).

🧾 Logging Pipeline

File logging is enabled by default with automatic rotation and retention cleanup (3 days by default):

"logging": {
  "enabled": true,
  "dir": "~/.clawgo/logs",
  "filename": "clawgo.log",
  "max_size_mb": 20,
  "retention_days": 3
}

Structured logging keys are now unified across channels and gateway:

• channel
• chat_id
• sender_id
• preview
• error
• message_content_length
• assistant_content_length
• user_response_content_length
• fetched_content_length
• output_content_length
• transcript_length

Constants are centralized in pkg/logger/fields.go.

🛡️ Sentinel & Risk Protection

Sentinel periodically checks critical runtime resources (config, memory, log directories), with optional auto-heal and notify:

"sentinel": {
  "enabled": true,
  "interval_sec": 60,
  "auto_heal": true,
  "notify_channel": "",
  "notify_chat_id": ""
}

Cron scheduling runtime strategy is configurable (and hot-reloadable):

"cron": {
  "min_sleep_sec": 1,
  "max_sleep_sec": 30,
  "retry_backoff_base_sec": 30,
  "retry_backoff_max_sec": 1800,
  "max_consecutive_failure_retries": 5,
  "max_workers": 4
}

Shell risk gate is enabled by default. Destructive commands are blocked unless explicitly forced:

"tools": {
  "shell": {
    "risk": {
      "enabled": true,
      "allow_destructive": false,
      "require_dry_run": true,
      "require_force_flag": true
    }
  }
}

🤖 Multi-Agent Orchestration (Pipeline)

Task protocol is standardized with: role + goal + depends_on + shared_state.

Available tools:

• pipeline_create: create a task graph
• pipeline_status: inspect pipeline status
• pipeline_state_set: write shared state
• pipeline_dispatch: dispatch currently runnable tasks
• spawn: supports pipeline_id/task_id/role

Channel commands:

/pipeline list
/pipeline status <pipeline_id>
/pipeline ready <pipeline_id>

🧠 Memory & Context Indexing

• memory_search: structured indexing (inverted index + cache)
• Memory layers: profile / project / procedures / recent notes
• Automatically Compacted Context: when a session gets long, it auto-summarizes and trims history

"memory": {
  "layered": true,
  "recent_days": 3,
  "layers": {
    "profile": true,
    "project": true,
    "procedures": true
  }
}

Automatic context compaction config:

"agents": {
  "defaults": {
    "context_compaction": {
      "enabled": true,
      "trigger_messages": 60,
      "keep_recent_messages": 20,
      "max_summary_chars": 6000,
      "max_transcript_chars": 20000
    }
  }
}

Hot-update examples:

clawgo config set agents.defaults.context_compaction.enabled true
clawgo config set agents.defaults.context_compaction.trigger_messages 80
clawgo config set agents.defaults.context_compaction.keep_recent_messages 24

🗺️ Repo-Map & Atomic Skills

• repo_map: build/query repository map before deep file reads
• skill_exec: execute skills/<name>/scripts/* atomically to keep gateway slim

📦 Migration & Skills

ClawGo now integrates all core extended capabilities from the original OpenClaw:

• coding-agent: Autonomous programming using Codex/Claude Code.
• github: Deep integration with gh CLI for managing issues, PRs, and CI status.
• context7: Intelligent context search for codebases and documentation.

🛠️ Installation (Linux Only)

Build from Source

cd clawgo
make build
make install

📜 License

MIT License. 🦐