Overview
When your AI assistant is connected to ACCELQ through MCP, it works with your projects using a set of ACCELQ tools: understanding your application, creating tests, running them, reporting on results and analyzing trends across runs. This article lists those tools and what each one does.
You don't call these tools yourself. Ask in plain language and the assistant picks the right one. Knowing what exists helps you ask for the right things. For ideas, see Example requests at the end of this article.
The assistant uses these tools as you, with your own ACCELQ permissions. It can only do what you can already do in ACCELQ.
Before you start: Connect your assistant to ACCELQ. See Connect an AI assistant to ACCELQ (MCP).
Projects and your application
The assistant uses these to find the right project and understand your application before writing tests. You don't need to ask for them.
| Tool | What it does |
|---|---|
list_projects |
Lists the ACCELQ projects you can access. Start here if you're not sure of a project's name. |
Writing tests
The assistant first proposes a test plan for your review. Tests are created once you approve it.
| Tool | What it does |
|---|---|
create_tests_from_plan |
Creates the tests from a test plan you have reviewed and approved. Large plans are built in the background. |
check_progress |
Reports progress on background work, such as test creation. |
stop_work |
Stops background work that is still going. Tests already created stay in ACCELQ. |
Finding what to run
| Tool | What it does |
|---|---|
resolve_entities |
Turns a scenario or test suite name you mention, such as "the login scenario", into the exact item in ACCELQ. If nothing matches confidently, the assistant asks you. |
get_entity_info |
Shows details of a scenario or test suite: its test cases, the values it needs at run time, and a short summary. |
get_project_config |
Lists a project's environments, execution parameters, app variants and members. For example, it answers "which environments do we have?" and finds the right teammate when you say "notify John" in a run request. |
Agents
| Tool | What it does |
|---|---|
list_agents |
Lists the execution agents you can run on, when each was last online, how busy it is, and which providers it uses. |
get_agent_capabilities |
Shows the browsers and mobile devices a specific agent offers. |
Running tests
| Tool | What it does |
|---|---|
get_run_plan |
Prepares a run and shows you what will run and where (environment, agent, browser, variant and run options) before anything starts. Nothing is started by this step. |
run_tests |
Starts the run you confirmed and gives you the job ID and a link. |
abort_job |
Stops a job that is running. Test cases that already finished keep their results. |
How run settings are chosen
What to run is always your choice. Other settings are taken from the first of these that applies:
- What you say in the chat
- Your last run in that project
- ACCELQ's defaults
You can ask for run options in plain language, for example:
- "keep the browser open"
- "with accessibility testing"
- "comprehensive screenshots and verbose logging"
- "page timeout 120 seconds"
Checking results
| Tool | What it does |
|---|---|
get_jobs |
Shows how specific jobs are doing: waiting, running or finished, with passed, failed and not-run counts. |
search_jobs |
Finds jobs from a description such as "last week's nightly regression runs on Chrome". |
Release readiness and trend analysis
Look across many runs at once to judge release readiness, spot flaky tests and new regressions, and see how results differ by browser and environment.
| Tool | What it does |
|---|---|
analyze_job_trends |
Analyzes a set of jobs: overall verdict, failing tests (persistent, flaky, regression or recovered), breakdowns by browser and environment, and linked defects. Good for "are we ready to release?" |
Example requests
Writing tests
- "Here are the checkout screens and our requirements doc. Create tests for guest checkout in project ECOM."
- "Create tests for the new password reset flow. Screenshots attached."
- "Create tests for PAY-1423 and PAY-1427 in project ECOM." (requires Jira Connector)
- "Create regression tests for the stories in the current sprint of the PAY board." (requires Jira Connector)
- "How is the test creation going?"
- "Stop creating the rest."
Note: Examples that mention Jira issues assume your assistant is also connected to Jira.
Running tests
- "Run the Login with valid credentials scenario."
- "Run the Smoke Regression suite on Firefox in the QA environment."
- "Run the Checkout suite in Staging and notify John."
- "Run T-4 on Local_agent with accessibility testing and a page timeout of 60 seconds."
- "Which agents can I run on right now?"
- "Stop job 5150."
Checking results
- "How is the run I just started doing?"
- "Is job 5150 finished yet?"
- "How many tests have passed and failed so far?"
- "Show me last week's nightly regression runs."
Release readiness and trend analysis
- "How did last week's regression runs go? Are we ready to release?"
- "Which tests are flaky across the last 10 nightly runs?"
- "Which failures are new regressions since last week, and which have recovered?"
- "Is anything failing only on Safari or only in the Staging environment?"
- "Which failing tests already have defects linked?"
Comments
0 comments
Please sign in to leave a comment.