Getting started¶

Installation¶

pip install robotframework-lynqa

Not on PyPI yet?

At this early stage the package may not be published on PyPI yet. In the meantime you can install it from source:

pip install git+https://github.com/petit-robot/robotframework-lynqa.git

Your first scenario¶

Import the library and write your test cases with the Given/When/Then keywords, each taking a single natural-language step:

*** Settings ***
Library    robotframework_lynqa.LynqaLibrary    api_key=%{LYNQA_API_KEY}

*** Variables ***
${LYNQA_URL}    https://lemonde.fr

*** Test Cases ***
Search An Article
    Given    the website is open
    When     I search an article on "france ia"
    Then     there is an article on "IA agentique"

The scenario is captured and submitted to Lynqa, which runs it against the site at ${LYNQA_URL} and reports each step's result back to Robot Framework.

Reading the Robot log¶

Beyond the pass/fail verdict, the library embeds Lynqa's execution evidence directly in the standard Robot Framework log (log.html). Under each Given/When/Then keyword you get a step-by-step account of what the agent actually did in the browser:

Commands done by the agent: every low-level action Lynqa performed to carry out the step. A command that failed is logged at ERROR level so it stands out in red.
Assertion results: each check is logged as an Assertion: "..." line, at INFO level when the assertion holds and at ERROR level when it fails. When a step fails, the reason reported by Lynqa is logged as a Verdict: line.
Screenshots: the page state is captured as an inline screenshot at each stage. Screenshots are embedded in a collapsible screenshot block, so you can expand only the ones you care about.

The Lynqa evidence embedded in the Robot Framework log

Configuration variables¶

Variable	Required	Description
`${LYNQA_URL}`	Yes	URL of the web application under test.
`${LYNQA_LANGUAGE}`	No	Language of the user running the scenario.
`${LYNQA_DATETIME}`	No	Date and time of the run (defaults to the current local date and time).
`&{LYNQA_SECRETS}`	No	Mapping of secret name to value (e.g. `login=superu password=TrèsS3cr3t`).

Providing your API key¶

The library authenticates with the api_key argument. Keep the key out of your test data by reading it from an environment variable with Robot Framework's %{ENV_VAR} syntax, as shown above:

export LYNQA_API_KEY="your-api-key"
robot my_tests.robot

Create a key from the Lynqa integration page.

Execution timeout¶

A Lynqa run is asynchronous: once a scenario is submitted, the library polls Lynqa until the run reaches a final status. By default it waits up to 1 hour before failing the test with a timeout error.

To change this behaviour, use Robot Framework's timeout feature, the Test Timeout setting in *** Settings ***, or a [Timeout] on an individual test case, which caps how long a test, including this wait, is allowed to run:

*** Settings ***
Library         robotframework_lynqa.LynqaLibrary    api_key=%{LYNQA_API_KEY}
Test Timeout    30 minutes

Next steps¶

Head to the Keyword reference for the full details of the Given/When/Then keywords and how they drive Lynqa.