/$$$$$$ /$$ /$$ /$$ /$$$$$$ /$$
/$$__ $$ | $$ | $$ | $$ /$$__ $$ | $$
| $$ \__/ /$$ /$$ /$$$$$$$ /$$$$$$ | $$$$$$$ /$$$$$$$| $$ /$$$$$$ /$$ /$$ /$$ | $$ \__/ /$$$$$$ /$$$$$$ /$$$$$$ /$$$$$$ /$$$$$$$ /$$$$$$
| $$$$$$ | $$ | $$| $$__ $$|_ $$_/ | $$__ $$ /$$_____/| $$ |____ $$| $$ | $$ | $$ | $$ /$$__ $$ |____ $$ /$$__ $$ /$$__ $$| $$__ $$|_ $$_/
\____ $$| $$ | $$| $$ \ $$ | $$ | $$ \ $$| $$ | $$ /$$$$$$$| $$ | $$ | $$ | $$ | $$ \ $$ /$$$$$$$| $$ \ $$| $$$$$$$$| $$ \ $$ | $$
/$$ \ $$| $$ | $$| $$ | $$ | $$ /$$| $$ | $$| $$ | $$ /$$__ $$| $$ | $$ | $$ | $$ $$| $$ | $$ /$$__ $$| $$ | $$| $$_____/| $$ | $$ | $$ /$$
| $$$$$$/| $$$$$$$| $$ | $$ | $$$$/| $$ | $$| $$$$$$$| $$| $$$$$$$| $$$$$/$$$$/ | $$$$$$/| $$$$$$/| $$$$$$$| $$$$$$$| $$$$$$$| $$ | $$ | $$$$/
\______/ \____ $$|__/ |__/ \___/ |__/ |__/ \_______/|__/ \_______/ \_____/\___/ \______/ \______/ \\_______/ \____ $$ \_______/|__/ |__/ \___/
/$$ | $$ /$$ \ $$
| $$$$$$/ | $$$$$$/
\______/ \______/
Your personal AI agent with full server control, 50 tools, 8 LLM providers, 1000+ app integrations, and an interactive CLI dashboard.
git clone https://github.com/truehannan/synthclaw-coagent.git
cd synthclaw-coagent
npm install && npm run link
synthclaw setup # interactive wizard
synthclaw start # starts the agent
synthclaw agent # interactive dashboard- Runs on a $6/mo VPS, talks to you via Telegram, WhatsApp, or CLI
- Executes shell commands, manages files, deploys services, calls APIs
- 50 built-in tools — shell, file I/O, HTTP, systemd, search, code execution, reminders
- 8 LLM providers — DigitalOcean, OpenAI, Anthropic, Google, NVIDIA, HuggingFace, OpenRouter, GitHub
- 1000+ app integrations via Composio (Gmail, Slack, GitHub, Notion, etc.)
- Dynamic API registration — give it a credential, it auto-detects the service and starts calling it
- MCP support — paste MCP JSON configs and the agent can use those tool servers
- Skills system — upload Claude-format skills (.zip) or install by name
- In-memory code execution — runs Node.js/Python/Bash without writing files
- Auto-resumes on failure — task state saved to SQLite, picks up where it left off
| Command | Description |
|---|---|
synthclaw setup |
Interactive wizard (edit mode if already configured) |
synthclaw start |
Start the agent (auto-creates venv, installs deps) |
synthclaw stop |
Stop the agent |
synthclaw agent |
Interactive dashboard — chat REPL with system meters |
synthclaw agent "task" |
Autonomous execution mode |
synthclaw status |
Show agent status + health |
synthclaw logs -f |
Tail logs |
synthclaw model <name> |
Switch LLM model |
synthclaw models |
List all available models |
synthclaw deploy |
Deploy to remote VPS |
| Command | Description |
|---|---|
/start |
Register as owner |
/model, /models |
Switch/list models |
/run <cmd> |
Execute shell command |
/plan <task> |
Plan without executing |
/agent <task> |
Autonomous execution |
/skills |
Manage skills |
/apis |
Manage registered APIs |
/connectors |
Composio app connections (1000+ apps) |
/mcp |
Manage MCP servers |
/memory, /creds |
Facts and credentials |
/usage |
Token usage & costs |
| Provider | Prefix | Models |
|---|---|---|
| DigitalOcean | (default) | Llama 3.3, Mistral, DeepSeek, Qwen, Claude, GPT-4o |
| OpenAI | openai- |
GPT-4o, GPT-4.1, o3 |
| Anthropic | anthropic- |
Claude Sonnet 4, Opus 4 |
google: |
Gemini 2.5 Flash/Pro | |
| NVIDIA | nvidia: |
Llama 405B, Nemotron, DeepSeek R1 |
| HuggingFace | hf: |
Llama, Qwen, Mistral, Phi-4 |
| OpenRouter | openrouter: |
Any model via proxy |
| GitHub | github: |
GPT-4o, Llama, Mistral |
Give the agent a credential — it auto-detects the service and registers it:
You: "Store my stripe key: sk_live_xxx"
Agent: Stored + auto-registered Stripe API (9 endpoints)
Later: "Show my charges" → agent calls Stripe API directly
18 services auto-detected: Stripe, Vercel, Cloudflare, GitHub, Notion, Airtable, etc.
/connectors connect github → OAuth link
/connectors connect gmail → OAuth link
Then the agent can send emails, create issues, post to Slack — all via natural language.
/mcp add {"mcpServers":{"filesystem":{"command":"npx","args":["-y","@modelcontextprotocol/server-filesystem","."]}}}
The agent runs code directly without writing files:
exec_code(code="console.log(Math.PI * 5**2)", lang="node")
exec_code(code="import json; print(json.dumps({'x': 42}))", lang="python")
/skills install coding-agent # search + install from GitHub
/skills install https://url/skill.zip # direct URL
Or send a .zip file to the bot — auto-installs if it contains SKILL.md.
synthclaw agentLaunches a terminal dashboard with:
- System gauges (CPU/MEM/DISK)
- Server info (IP, uptime, model)
- Chat REPL with
/command autocomplete - Red theme throughout
The wizard detects if you've already configured:
- First run: full setup (storage, interface, credentials, model, limits)
- Re-run: edit mode (Enter to keep current values)
Options include:
- Storage: Local SQLite or Cloudflare D1 + R2
- RPM limiting: cap requests per minute to control costs
- Composio: optional 1000+ app integration key
┌──────────────┐ ┌──────────────┐ ┌──────────────┐
│ Telegram │ │ WhatsApp │ │ CLI Dashboard│
└──────┬───────┘ └──────┬───────┘ └──────┬───────┘
└────────────┬────┘───────────────────┘
│
┌──────▼──────┐
│ Agent Core │ ← 8 LLM providers + tool loop
└──────┬──────┘
│
┌──────────────┼──────────────┐
│ │ │
┌────▼────┐ ┌─────▼─────┐ ┌────▼────┐
│ 50 Tools│ │ Memory │ │Composio │
│exec_code│ │ SQLite │ │ MCP │
│api_call │ │ +Fernet │ │ Skills │
└─────────┘ └───────────┘ └─────────┘
Source Available — Non-Commercial. Free for personal use.
Made by @truehannan