ComfyUI Telegram Suite

Implement Telegram into your ComfyUI workflows.

Nodes

The main nodes:
<img src="https://github.com/SwissCore92/comfyui-telegram-suite/blob/master/screenshots/main_nodes.png" alt="screenshots/main_nodes.png">

<details><summary>Telegram Bot </summary> This node loads your Telegram bot and (optionally) sets a default chat.

You can configure it via: ComfyUI/user/default/telegram-suite/config.json

</details> <details><summary>Send Message </summary> This node just sends a simple text message. </details> <details><summary>Send Image(s) </summary> This node sends one or more (up to 10) images.

• If the IMAGE input contains multiple images and group is set to True, they’ll be sent as a media group.
• If group is False, the images will be sent individually.
• If send_as_file is True, the images will be sent as files instead of inline media.

Note:
Only the message(_id) of the last sent image will be returned to the output.

</details> <details><summary>Send Video </summary> This node sends a video file.

• The video input must be of type VHS_FILENAMES (e.g., from the Filenames output of the Video Combine node in the Video Helper Suite).

The video can be sent as a regular video, an animation, or a file.

</details> <details><summary>Send Audio </summary> This node sends an audio file.

• Can be sent as an audio message, voice message, or file.

</details> <details><summary>Send Chat Action </summary> This node sends chat actions like “typing...”, “uploading X...”, or “recording X...”.

⚠️ This is not an output node, so the trigger passthrough is required for this node to work.

</details>

Additional nodes include message editing, experimental features, and various type converters (see Triggers).

Installation

Note: Requires ffmpeg!

Step 1:

Install via ComfyUI Manager (and skip to Step 2) or execute the following commands:

⚠️ Ensure your ComfyUI virtual environment is activated and you're in the ComfyUI/custom_nodes directory.

git clone https://github.com/SwissCore92/comfyui-telegram-suite.git
cd comfyui-telegram-suite
pip install -r requirements.txt

Step 2:

Restart ComfyUI.

Step 3:

Add your bot(s) and chat(s) to the config file.

• Open: ComfyUI/user/default/telegram-suite/config.json.
• Add your bot token(s) under "bots".
• Add your chat ID(s) under "chats".

Example:

{
    "bots": {
        "MyCoolBot": "123456:ABC-DEF1234ghIkl-zyx57W2v1u123ew11",
        "MyOtherCoolBot": "654321:CBA-DEF1234ghIkl-zyx57W2v1u123ew11"
    },
    "chats": {
        "MyPrivateChat": 567891234,
        "MyGroupChat": -1012345678
    }
}

Use any string as the key for "bots" and "chats" — I like to use the Telegram @username for clarity.

Step 4:

Restart ComfyUI again — and you're good to go!

Triggers

The optional trigger inputs/outputs are used to enforce execution order in your workflow.

ComfyUI runs by evaluating output nodes and working backward to resolve dependencies. I like to think of it as the inputs "pulling" the values they need from connected outputs.

The trigger passthrough ensures a node executes at a specific point during the workflow. Here's an example using F5-TTS:

<img src="https://github.com/SwissCore92/comfyui-telegram-suite/blob/master/screenshots/trigger_example_tts.png" alt="screenshots/trigger_example_tts.png">

This flow sends the recording_voice chat action before generating the audio with the F5-TTS node. Once audio is generated, it's sent to the chat.

The seed is required by the F5-TTS Audio node, so the Send Chat Action node must execute first.

You can use almost any type as a trigger. However, since ComfyUI has strict type checking, you'll need to:

• Convert the signal to ANY before feeding it into the trigger input.
• Convert it back to the original type after the trigger output.

Yes, it's a bit clunky — but it’s the only reliable way I’ve found to control execution order. That’s also why the converter category has so many nodes.

To Do

• [ ] Improve documentation
• [x] Add Edit Message Video node
• [x] Add Edit Message Audio node
• [ ] Add tooltips and descriptions.
• [ ] Wait for feedback to refine this list