Skip to content

Telegram & Discord Commands

All commands work identically on both Telegram and Discord. Responses route back to the originating transport.

General

Command Description
/ping Connectivity check — returns pong
/status System stats (uptime, memory, active skills)
/help Full command list

Skills

Command Description
/skills List all registered skills (core + dynamic)

Skill management

Loading, unloading, and reloading skills is Admin UI only (2FA enforced).

Workflows

Command Description
/workflows List all workflows with status and schedule
/run <id\|name> Execute a workflow on demand (supports 2FA gating)
/runs List active (running) workflows
/cancel <run_id> Cancel a running workflow
/rate <run_id> View/rate workflow step outcomes (thumbs up/down)

Search & Research

Command Description
/search <query> Web search with LLM synthesis
/search --tier <T> Save default search tier
/research <query> Deep research with memory context
/research --tier <T> Save default research tier
/web <url> [question] Fetch URL and summarize or answer
/web --tier <T> Save default web browse tier

GitHub

Command Description
/github pr <owner/repo> [number] Review PR for security issues
/github commit <owner/repo> <sha> Review commit for security issues

Shell

Command Description
/shell <command> Execute whitelisted OS command (2FA-gated)

Code Generation

Command Description
/coder <goal> Generate a dynamic skill from description

Google Services

Command Description
/connect google Initiate Google OAuth flow
/tasks List Google Tasks
/task add <title> Create a new task
/tasklists List task lists

Browser Automation

Command Description
/record <url> Start browser recording
/record stop <session_id> Stop recording, save as resource
/replay <id> Replay a saved automation
/automate <url> Headless scrape/screenshot

LLM

Command Description
/llm Provider status and daily usage

2FA

Command Description
/setup 2fa Generate TOTP secret and QR code
/confirm 2fa <code> Confirm 2FA setup
/disable 2fa Disable 2FA

Free Text

Any message that doesn't match a command is routed to the conversational skill for LLM-powered conversation with memory context.