AI Features

reqable can use AI to generate mock data and refine existing responses, reducing the work required to write JSON, adjust fields, and maintain test fixtures.

Configuration

Open Settings → AI model configuration and configure:

1. An API protocol: OpenAI, Anthropic, or Gemini.
2. The Base URL.
3. The API Key.
4. The model name.

TIP

Many providers, including DeepSeek, GLM, and MiniMax, expose OpenAI-compatible APIs. Select OpenAI and enter the Base URL and API Key supplied by your provider.

Entry points

Generate mocks during YApi import

After selecting interfaces from YApi, choose AI generation. reqable sends the endpoint path, method, request schema, response schema, and description as context.

This is useful when:

• The YApi mock response has not been maintained.
• The response contains many fields and would be slow to write manually.
• You want to generate multiple mock rules in one operation.

Progress is shown while generation runs. A failed endpoint does not cancel successful items; all failures are summarized when the batch finishes.

Refine a response from the details view

Use Refine with AI in the response editor:

1. Enter an instruction such as “Reduce the list to three items” or “Convert amount fields to integers.”
2. The model streams an updated response.
3. reqable displays a diff after generation.
4. Accept or reject the result. The current mock is not overwritten automatically.

Demo

INFO

Mocks imported from YApi retain the endpoint schema as AI context, so later refinements usually stay closer to the API definition.

Relationship to normal mocks

AI assists with data creation but does not change the underlying mock workflow:

• Generated items remain normal mock rules.
• You can still edit the response, match strategy, and capability switches manually.
• Rules still support import, export, groups, enabling, and disabling.
• If you do not want to use AI, import the existing YApi mock or create a rule manually.

Recommendations

• Use YApi import plus AI generation when the endpoint definition is complete.
• Use direct YApi mock import when YApi already contains stable mock data.
• Use response refinement when a mock exists and only needs a focused change.