OpenAI 文本块
The text chunk builder for OpenAI messages
- content_item
The most boring node in the pack, on purpose
OpenAITextBlock wraps any string as an OAI_CONTENT_ITEM - a {"type": "text", "text": ...} part - so it can flow through the same content pipeline as encoded images and uploaded files. One input, one output, zero settings.
You won't reach for it if you're sending a single literal prompt - OpenAIChat's user_prompt covers that. It earns its keep when the text isn't a typed literal but a value arriving from elsewhere in the graph: a wildcard expansion, a metadata string, an OCR output, a transcription. Drop that dynamic string into text, and the content_item output plugs into an OpenAIChat content slot or sits beside an encoded image in an OpenAIContentConnector.
Inputs and output
text - a multiline string, default empty. content_item - the resulting OAI_CONTENT_ITEM.
That's all there is. No escaping, no templating, no merging - combine multiple texts upstream or run several blocks. If you want text plus image plus file in one message, text block + encoder/uploader + connector is the intended assembly.
Gotchas
- Sending the same text in
user_promptand in a text block duplicates it in the message. Pick one place. - An empty block produces an empty text part, which the Chat node skips as falsy - harmless, but keep the graph readable.
- Local-only: nothing is sent until the content item reaches
OpenAIChat, so a dormant block costs nothing.
Install
With the pack via ComfyUI Manager (search "comfyui-aichat") or:
cd ComfyUI/custom_nodes
git clone https://github.com/spawner1145/comfyui-aichat
Restart ComfyUI. No models to download; dependencies are just openai, httpx, and PyYAML. When you're building multi-part messages, a node this simple is exactly what you want between the parts that actually do something.
Inputs (1)
| Name | Type | Default | Description |
|---|---|---|---|
| text | STRING | — |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| content_item | OAI_CONTENT_ITEM | — |