Skip to content

Step 2: Prepare a prompt set

A prompt set groups several prompts dedicated to a given use of AI in SquashTM (today, classic test case generation). It dictates how the AI should reply: its role, the expected output format, and the variables injected from SquashTM.

Prerequisites

  • An account with administrator rights.
  • A SquashTM Ultimate 💎 license.

Option A: Use a default prompt set

SquashTM ships with two pre-configured prompt sets, one in French and one in English. Each set contains three ready-to-use prompts that produce test cases of increasing richness:

PromptWhat the generated test case contains
Standard generationName, description, prerequisites, and test steps.
Detailed generationSame, plus weight and reference.
Detailed generation with datasetsSame, plus datasets that allow parameterizing the test case.

These sets are identified by the "S" badge (System prompt set) in the management grid. They are read-only and cannot be deleted.

For a quick start, use a default prompt set

If this is your first setup, jump directly to step 3 and associate the default prompt set that matches your project's language. You can create a custom set later.

Option B: Create a custom prompt set

If you need to adjust the tone, the output format, or inject specific data, create your own set.

Access the management page

Go to Administration > Entities, then click the Prompt sets anchor Prompt sets icon.

Prompt set management table

Create the set

Click Add Add and fill in:

  • Name: unique, mandatory;
  • Usage: Classic test case generation (the only usage available today);
  • Description: optional.

Prompt set creation pop-up

Add one or more prompts to the set

Open the set by clicking its # or its Name, then add a prompt with Add.

A prompt is composed of three elements:

  • a system prompt template: defines the AI's role and the expected JSON output format;
  • a user prompt template: carries the request, enriched with the requirement data ({{requirement.name}}, {{requirement.description}}, etc.);
  • a context wrapper template (optional): controls how the linked requirements and attached documents are integrated into the message sent to the AI.

Add prompt pop-up

Handlebars syntax

The three templates use the Handlebars syntax ({{variable}}, {{#if}}…{{/if}}, {{#each}}…{{/each}}). For the full list of available placeholders and detailed examples, see Manage prompt sets.

Take inspiration from the default prompts

The full content of the three default prompts (system, user, context wrapper) is available on the reference page: Prompt examples for test case generation. Copy them as a starting point and adapt them to your needs.

What is the context wrapper used for?

The context refers to the additional information that the tester can add to their request at generation time: requirements linked to the source requirement and attached documents (specifications, mock-ups, etc.). The context wrapper template controls how this information is formatted and inserted into the message sent to the AI.

If this template is filled in, the Context & Sources panel will be offered to the tester during generation, allowing them to add requirements and documents as context.

If this template is left empty, the panel will not be displayed and the generation will rely solely on the source requirement.

What you have accomplished

  • You decided to use a default prompt set or created a custom set;
  • the chosen set contains at least one ready-to-use prompt;
  • this set is now available in each project's configuration.

Next step

Enable AI on a project.