@ironcode-ai/telegram
v1.20.7
Published
Telegram bot integration for ironcode. Send messages from Telegram to run an AI coding agent on your machine — tool calls, file edits, and responses stream back in real-time.
Readme
@ironcode-ai/telegram
Telegram bot integration for ironcode. Send messages from Telegram to run an AI coding agent on your machine — tool calls, file edits, and responses stream back in real-time.
Requirements
- Bun >= 1.0
ironcode-aiCLI installed and authenticated
Installation
1. Install ironcode-ai CLI
bun install -g ironcode-ai2. Authenticate a provider
ironcode auth login # GitHub Copilot (recommended)
ironcode auth login anthropic # Anthropic
ironcode auth login openai # OpenAIList available models:
ironcode models3. Create a Telegram Bot
- Open Telegram → search for @BotFather
- Send
/newbot→ follow prompts → copy the Bot Token
4. Install ironcode-telegram
bun install -g @ironcode-ai/telegram5. Configure
ironcode-telegram setup
# Bot Token (from @BotFather): ****
# Model [github-copilot/claude-sonnet-4.6]:
# ✅ Config saved to ~/.config/ironcode/telegram.jsonConfig is stored at ~/.config/ironcode/telegram.json:
{
"token": "your-bot-token",
"model": "github-copilot/claude-sonnet-4.6"
}6. Run
# cd into the repo you want the agent to work on
cd /path/to/your/project
ironcode-telegramThe agent runs with the current directory as its working directory, so it can read and edit files in your project.
Bot Commands
| Command | Description |
| ----------- | --------------------------------------------------------- |
| /sessions | List recent sessions — tap to switch |
| /new | Start a new session |
| /info | Show current session details (title, ID, file changes) |
| /init | Analyze the project and create an AGENTS.md config file |
| /diff | Show all file changes made in the current session |
| /start | Show help |
How It Works
You send a message
→ Bot shows 🤔 Thinking... while the agent starts up
→ Bot creates/resumes an ironcode session on your machine
→ Agent reads/writes files, runs bash, calls LLM
→ Each completed tool call is sent as a separate message
→ Text response is streamed live by editing the placeholder
→ Final response replaces the placeholder when doneEach chat (DM, group, or group thread) gets its own independent session. Use /sessions to switch between them.
Upload Support
The bot supports multiple file types for code analysis and generation:
📸 Images (Photos)
- Screenshots — UI bugs, design mockups, error messages
- Diagrams — Architecture diagrams, flowcharts, wireframes
- Code screenshots — OCR extraction and conversion
- Send as single image or multiple images as album
📄 Documents
- Code files —
.js,.ts,.py,.java, etc. - Text files —
.txt,.md,.json,.xml - PDFs — Technical docs, API specs
- Archives —
.zipfiles (extracted and analyzed) - Max file size: 20MB per file
🎤 Voice Messages
- Transcribed using Groq Whisper API
- Requires Groq API key in config
- Perfect for hands-free coding instructions
Usage:
- Send a photo/file/voice message (optionally with caption)
- Bot downloads and processes it
- AI agent analyzes and responds based on content
Examples:
- Send screenshot of error → "Fix this bug"
- Upload
.pyfile → "Add type hints and docstrings" - Send multiple images → "Compare these UI designs"
- Voice message → "Refactor the user auth module"
Session Management
- Automatic: Each chat automatically gets its own session on first message.
- Switch:
/sessionsshows the 10 most recent sessions with an inline keyboard. Tap any to resume it. - New:
/newclears the current slot — your next message starts a fresh session. - Persistent: Sessions and their file change history are preserved across bot restarts.
