Telegram Commands
Built-in slash commands for controlling the agent from Telegram.
| Command | Description |
|---|---|
/model | Show the current active model |
/model <name> | Switch to a specific model (disables auto-routing) |
/models | List all available models from the provider (active model marked) |
/auto | Re-enable automatic model routing based on query complexity |
/reload | Reload config.toml (with auto-restore from backup if broken) |
/restart | Full restart — exec new process (picks up new binary, config, MCP servers) |
/cost | Show token usage statistics (last 24h, 7d, top models) |
/tasks | List running and recent agent tasks for your session |
/cancel <id> | Cancel a running task by ID |
/help | Show list of available commands |
/start | Same as /help |
Model Switching
text
/model claude-3.5-sonnet
# Switches to claude-3.5-sonnet and disables auto-routing
/auto
# Re-enables automatic Fast/Primary/Smart routingModel Override
When you manually set a model with
/model, auto-routing is disabled until you send /auto.Config Reload
The /reload command reloads config.toml at runtime. If the config is invalid, aidaemon automatically restores from backup:
.toml.lastgood— last config that successfully completed an LLM call.toml.bak→.toml.bak.1→.toml.bak.2— 3-deep rotation