Skip to content

Prompts

Good prompts are worth keeping. Chatty the Lab lets you save one once and reuse it anywhere, instead of retyping it or hunting through old conversations.

The thing to know first: there is no separate prompt table. A saved prompt is a context in the Library, and the prompt picker reads that same list. Knowing this explains two things that are otherwise puzzling — why saved prompts appear grouped by category next to material that is not a prompt, and why they follow the workspace rather than you.

Contexts live in Settings → Library → Contexts. Each has a title, a category, optional tags, and the text itself. Give a prompt the category prompt and it groups with the others.

The Library's Contexts tab filtered to one saved prompt, grouped under the category prompt
The Contexts tab, filtered. A saved prompt is a context: title, category, tags, and the text. The heading above it is the category the entries are grouped by.

The PromptPicker appears wherever a graph exposes a parameter that is a prompt — in a chat’s parameter panel and in a node’s configuration. It lists what the Library holds, so a prompt saved once shows up in every one of those places.

  1. Open the parameter panel of a chat bound to a graph, or a node’s configuration in Studio.

  2. Where the parameter is a prompt, the picker lists your saved contexts. Pick one and it loads into the field.

  3. Edit it in place with the pencil if this run needs a variation — editing there changes the parameter, not the saved context.

Contexts belong to the active workspace, so a team shares one library everyone can reuse, and what you saved in your personal workspace is not visible from a team. The exception is the platform’s own documentation, which is global and shows up in every workspace’s Contexts tab.

Beyond your own library, you can configure external prompt search sources so the picker can pull in prompts from outside Chatty. These sources are set up in Settings and synced on demand.