Learns from every message
Every conversation feeds the LoRA training loop. More users means faster evolution — zero manual effort.
AUTO TRAININGConnect your self-evolving agent to Telegram and Discord in minutes. Your users chat. Your agent learns. Zero extra setup.
Deploy EvoClaw on Telegram, Discord, or both. Every conversation feeds the same learning loop.
Simple setup. Works on any device. Users just send messages — the agent responds, learns, and evolves automatically.
5 MINUTE SETUPFull server integration with slash commands, embed cards, and per-channel memory. Perfect for communities and dev teams.
SLASH COMMANDSEvery conversation feeds the LoRA training loop. More users means faster evolution — zero manual effort.
AUTO TRAININGOne EvoClaw instance. Both bots point to the same proxy. Skills learned on Telegram improve Discord responses too.
SHARED BRAINThe bot runs on your VPS. Your conversations never leave your infrastructure. Full privacy, full control.
SELF-HOSTED5 steps. Takes about 5 minutes. No coding required.
Make sure EvoClaw is installed and running on port 8080 before connecting your bot.
pip install evoclaw evoclaw init evoclaw start
curl http://localhost:8080/v1/models — should return a JSON response.BotFather is Telegram's official bot manager. Open Telegram and follow these steps:
@BotFather/newbotEvoClaw Agent — username must end in _bot, e.g. evoclaw_agent_bot1234567890:ABCdef... — keep it safe, never share publicly.The Telegram bot script is included in the EvoClaw repo. Pull the latest version.
cd ~/evoclaw/repo
git pull origin main
ls evoclaw_telegram.py # should exist
Install the required libraries, then run the bot inside a screen session so it stays alive 24/7 — even after you disconnect from SSH.
# Install dependencies pip install python-telegram-bot openai --break-system-packages # Run in background (stays alive after SSH disconnect) screen -dmS tg-bot bash -c "cd ~/evoclaw/repo && TELEGRAM_BOT_TOKEN='YOUR_TOKEN' python3 evoclaw_telegram.py" # Verify it's running screen -ls
Open Telegram, search for your bot's username, and send your first message.
/start/statsPRM score=0.xx | conv=N | skills=NMore features than Telegram — embed cards, slash commands, and per-channel conversation memory.
Go to discord.com/developers/applications and create a new application.
New ApplicationEvoClawBot → Reset TokenMESSAGE CONTENT INTENT and SERVER MEMBERS INTENT → Save Changes.Generate an invite URL and add the bot to your server.
OAuth2 → URL Generatorbothttps://discord.com/api/oauth2/authorize?client_id=YOUR_CLIENT_ID&permissions=277025508352&scope=bot
Run inside a screen session so the bot stays alive 24/7.
# Install dependencies pip install discord.py openai --break-system-packages # Run in background screen -dmS dc-bot bash -c "cd ~/evoclaw/repo && DISCORD_BOT_TOKEN='YOUR_TOKEN' python3 evoclaw_discord.py" # Verify it's running screen -ls
Open your Discord server and try these to confirm the bot is working.
@EvoClaw hello!/ask what is LoRA?!statsQuick fixes for the most common issues.
Check proxy is running: evoclaw status
Check screen is alive: screen -ls
Make sure your token is correct and not expired.
The proxy is not running. Start it: evoclaw start
Check if port 8080 is already in use: ss -tlnp | grep 8080
Always use screen to run the bot in the background. Never run it directly in an SSH terminal.
Command: screen -dmS tg-bot bash -c "..."
Enable in Developer Portal: Bot → Privileged Gateway Intents → MESSAGE CONTENT INTENT → ON → Save Changes.
# List all running screen sessions screen -ls # Expected output: tg-bot (Detached) ✅ — Telegram bot dc-bot (Detached) ✅ — Discord bot evoclaw-proxy (Detached) ✅ — EvoClaw proxy evoclaw-agent (Detached) ✅ — Agent runner # Check proxy status evoclaw status # View logs for a specific session screen -r tg-bot
Both bots take less than 10 minutes to deploy. Your agent starts learning from the very first message.