maleon17

Claude Ask — Development skill for Claude Code

Development community

Standalone Claude Jarvis Telegram userbot and ClaudeAsk backend.

How to install Claude Ask

This entry records only its repository, not the path inside it, so there is no exact command to give. Open maleon17/claude-ask and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What Claude Ask does

Standalone Claude Jarvis Telegram userbot and ClaudeAsk backend.

Alternatives in Development

Telegram — Show Telegram bot status and manage global session 1.3k ★
Agentlas Connect — Connect Agentlas agents or teams to Telegram 1.1k ★
Claude Telegram Relay — Minimal pattern for running Claude Code as an always-on Telegram bot 326 ★

README

claude-ask

Part of **[telegram-ai](https://github.com/maleon17/telegram-ai)** — Claude/Codex ↔ Telegram, four ways.

A standalone Telegram userbot/backend for ClaudeAsk. It lets the owner invoke Jarvis with the `.ask` command in any Telegram chat and perform real actions through their own Telethon session. This is not the Bot API bridge and not codex-ask.

Repository:

Components

claude_ask.py — the main loadable ClaudeAsk module;
claude_watcher.py — the Claude Code backend with persistent sessions;
mcp_telegram_tools.py — MCP tools for Telegram account actions;
cmd_queue.py — the shared HTTP queue relay for .ask and .xask;
setup.sh and service examples — backend infrastructure setup.

Commands

ClaudeAsk uses `.ask`, `.search`, `.translate`, and `.new`. Triggers are stored in the `ClaudeAsk` namespace. CodexAsk uses a separate history namespace and `x`-prefixed commands, but deliberately reads the same trigger namespace; the incoming-message watcher stays a single one, so triggers aren't duplicated.

Architecture

The userbot module sends a request to `/ask`, the worker processes it through Claude Code and returns a progress/answer stream via `/tmp/jarvisask_ask_*`. MCP calls land in `/tmp/jarvisask_tool_queue`; the module holding a live Telethon session executes the action and returns the result. Every action gets a real tool result first, and only then is it reflected in the reply to the user.

The queue relay is shared infrastructure transport. If `jarvis-ask-cmd-queue.service` is already running on the host, don't start a second copy on the same ports/directories.

Backend installation

Requirements: Linux, Python 3.10+, the Claude Code CLI authenticated via `claude auth login --claudeai`, systemd (for running as a service). The userbot itself must be installed separately — this project targets the [Heroku](https://github.com/coddrago/Heroku) userbot; follow its own README for in