Two different tools, one key (set it up first): chat is you talking to the model about your flow; ai-action is your flow talking to the model while it runs.
The Chat tab
Select a node and open the Chat tab in the inspector. The conversation has your flow as context, so questions can be direct:
- “Why did this assertion fail?”
- “Write a
checkthat verifies the response body is a non-empty array of users.” - “What does this 422 response mean?”
Enter sends; Shift+Enter makes a newline. If no key is configured, the chat points you to Settings.
The ai-action node
An ai-action node makes a model call one step of the flow — with the run’s context
(response data, variables) available to the prompt:
- id: summarize_failures
type: ai-action
config:
prompt: "Summarize which assertions failed and suggest a fix."
output: aiSummary # optional — where the reply lands
next: nullThe reply is stored in variables under the output name, readable by every later node —
same as a captured value. On the graph the node shows a
thinking state while the model responds.
Good uses: summarizing a batch of results at the end of a run, classifying a free-text response before a condition branches on it, generating varied test data mid-flow.
In the app, ai-action nodes use the same key as chat. They also run headlessly: the
CLI and MCP server pick up ANTHROPIC_API_KEY from the
environment (ANTHROPIC_MODEL optionally picks the model), so an ai-action flow works in
CI too.
A caution worth stating: model output isn’t deterministic. For pass/fail decisions in CI,
prefer assertion nodes with concrete check expressions, and let ai-action handle the
judgment-and-prose work around them.