Skip to main content

Thunders MCP available tools

Written by Ines

Everything the Thunders MCP server exposes: 53 tools. If your organization role is Read-only, your assistant sees only the tools marked Read.

Projects and apps

Tool

Access

What it does

Parameters

get_all_projects

Read

List every project you can access.

filter, orderBy, first, all optional. Filter and sort on id, name, url.

get_apps_by_project_id

Read

List the apps configured in a project.

projectId

Test cases

Tool

Access

What it does

Parameters

get_all_test_cases

Read

List test cases with their labels, test sets and steps. The response is large: to find a test case, use search_test_cases.

projectId (strongly recommended), first, orderBy, fields, all optional. Use fields to keep only the fields you need.

search_test_cases

Read

Find the test cases of a project by text, labels and status. Returns a short summary of each match, the most recently updated first. Use get_test_case_by_id to read the steps of a result.

projectId. Optional: query (text in the name or the description), labels (a test case must carry every one of them), status, fields, first.

get_test_case_by_id

Read

Fetch one test case, including its steps.

id. Optional: fields.

get_test_cases_by_label

Read

List every test case in a project carrying a given label.

projectId, labelName

create_test_case

Write

Create an empty test case.

projectId. Optional: name, description, executionPlatform (Web or Mobile), reusedTests.

clone_test_case

Write

Duplicate a test case with its steps, settings and test sets. The copy is a Draft named "(Copy)".

id

add_label_to_test_case

Write

Label a test case. Creates the label if the project does not have it yet.

testCaseId, projectId, labelName

delete_test_case

Destructive

Delete one or more test cases. A test case reused by another test case is not deleted: it comes back under reusedIds and stays in place, so check deletedIds to confirm what actually went. If any id belongs to another organization the whole call is rejected.

testCaseIds (a list, so one call can remove several)

queue_test_cases

Destructive

Queue test cases for execution. The response gives the testRunIds of the new runs: pass them to wait_for_test_runs or get_test_run_status.

testCaseIds, projectId, environmentId, personaId. Optional: browserSettings, maxParallelism, maxRetries, alwaysRerun, executionMode (parallel or sequential), failureBehavior (continue or stop), variables, testAssetReferences.

Authoring guide

Tool

Access

What it does

Parameters

get_authoring_guide

Read

How to write Thunders test steps: the step types and their JSON, variables, the credit cost of each way to write steps, how a run is judged, and how to fix a failed run. Your assistant reads it once, before it writes or fixes steps.

Optional: platform (Web, the default, or Mobile). The step types differ by platform.

Test steps

Tool

Access

What it does

Parameters

generate_test_steps

Write

Turn a plain-language scenario into executable steps on a test case. By default it adds steps and changes no existing step. With toolset set to browser (or mobile_browser for a Mobile App), it opens the live app and uses the labels that the page shows. A call that browses costs 0.20 credits, a call that does not browse costs 0.05 credits.

testCaseId. Optional: scenario, toolset, mode (Append, the default, or Replace to let it edit or remove existing steps), stage (Main or Teardown), testCaseUpdatedAt for conflict detection, personaId, projectId, language (for example en, fr), attachmentIds, mobileBrowsePlatform (android or ios).

insert_step

Write

Insert one new step at a given position (0 is first). Thunders assigns the step id. Any step type is supported, including Execute JavaScript: get_authoring_guide gives the JSON of each type.

testCaseId, step, position. Optional: knownUpdatedAt for conflict detection.

update_test_step

Write

Rewrite one existing step. Any step type is supported, including Execute JavaScript.

testCaseId, step. Optional: knownUpdatedAt for conflict detection.

delete_step

Destructive

Delete one step. Every step nested under it is deleted too, and a Conditional step that jumped to a deleted step stops the run instead.

testCaseId, stepId. Optional: knownUpdatedAt for conflict detection.

set_test_steps

Destructive

Replace the whole step list of a test case in one write. Keep the id of every step you keep, and leave it out on new steps. Steps you leave out are deleted.

testCaseId, steps (the complete ordered list), knownUpdatedAt (required: the last known updatedAt of the test case).

Test sets

Tool

Access

What it does

Parameters

get_test_sets

Read

List the active test sets in a project (id and name), excluding deleted sets and plans. Use it to resolve a test set name to its id.

projectId

create_test_set

Write

Create a test set, optionally seeded with existing test cases.

projectId. Optional: name, testCases.

add_test_case_to_test_set

Write

Add a test case to a test set. Does nothing if it is already a member.

testCaseId, plus exactly one of testSetId or testSetName. A name must match exactly one test set in the test case's own project.

remove_test_case_from_test_set

Destructive

Remove a test case from a test set. This changes membership only: it does not delete the test case or the test set.

testCaseId, plus exactly one of testSetId or testSetName.

Schedules

Tool

Access

What it does

Parameters

get_schedules_by_test_set

Read

List the schedules of a test set, with their rule, status and next run.

testSetId

create_schedule

Write

Run a test set on a recurring rule. Each run uses the test cases that the set holds at run time, so a test case added later is included.

testSetId, frequency (Hourly, Daily, Weekdays, Weekly or Monthly), timezone, environmentId, personaId. Depending on the frequency: minutePastHour, timeOfDay, daysOfWeek, daysOfMonth. Optional: name, executionSettings.

update_schedule

Write

Change the rule, environment, persona or execution settings of a schedule. Send the whole rule: a field you leave out is not kept. The test set never changes.

scheduleId, plus the same fields as create_schedule except testSetId.

pause_schedule

Write

Stop a schedule from running. Past runs stay, and you can resume it later.

scheduleId

resume_schedule

Write

Resume a paused schedule. The next run is computed from now, so missed runs do not run. A schedule that Thunders paused stays paused until you fix the problem.

scheduleId

duplicate_schedule

Write

Copy a schedule onto the same test set. The copy starts paused, so it cannot run before you check it.

scheduleId

run_now

Destructive

Run the test set of a schedule now, with its saved environment, persona and execution settings. Works on a paused schedule, and does not move the next scheduled run. Refused while a run of the same schedule is in progress, or when your organization has no credits left.

scheduleId

delete_schedule

Destructive

Delete a schedule permanently. Its past runs stay.

scheduleId

Test assets

Tool

Access

What it does

Parameters

list_test_assets

Read

List the uploaded data files available in a project.

projectId

upload_test_asset

Write

Upload a data file (CSV, JSON, PDF, image and so on, max 20 MB). Reference it from a test step by its file variable, for example [FILE_CUSTOMERS_CSV].

projectId, fileName (with extension), contentBase64. Optional: testId to associate it with a test case.

Step quality suggestions

Thunders reviews authored steps against its authoring best practices before you run them.

Tool

Access

What it does

Parameters

get_test_case_suggestions

Write

List pre-run suggestions for a test case, each with a score, the rule violated, and a rewritten step. It counts as a write because the first call generates and stores the evaluation.

testCaseId

apply_test_case_suggestion

Destructive

Replace the step's text with the suggested rewrite.

testCaseId, suggestionId

dismiss_test_case_suggestion

Write

Hide a suggestion. Leaves the step untouched.

testCaseId, suggestionId

Test runs

Tool

Access

What it does

Parameters

get_test_runs_by_project_id

Read

List runs for a project, newest first, one page at a time. A test set execution shows as one row.

projectId. Optional: filter, orderBy, first, after, label. Pass the endCursor of a page as after to read the next page.

get_all_test_runs

Read

List runs across everything you can access, newest first, one page at a time.

filter, orderBy, first, after, label, all optional. Filter on id, test_case_id, test_set_id, status, run_at.

get_test_run_by_id

Read

Fetch one run's record, with its status name and, for a failed run, the failed step and the reason. The step recap is left out unless you ask for it.

id. Optional: include (recap, context).

get_test_run_context

Read

The full debugging picture for one run: scenario, authored steps, environment, persona, variables (secrets redacted), per-step recaps with bug reports, and console logs. This is the one to use when investigating a failure.

id

get_test_runs_by_ids

Read

Fetch up to 50 runs in one call, with the same fields as get_test_run_by_id.

ids. Optional: include (recap).

get_test_run_status

Read

Check the status of up to 50 runs in one call, without step recaps.

ids

wait_for_test_runs

Read

Wait until every given run finishes, then return their status. It waits 220 seconds at most: when timedOut is true, call it again.

ids. Optional: timeoutSec (default 120, maximum 220), pollSec (default 15).

get_run_evidence

Read

A compact view of one run, to judge whether it proves anything: each step with its status, reasoning and bug reports, and the hosts the run reached. It flags an assertion that may prove nothing, so you can check its screenshot.

runId

get_step_screenshot

Read

The screenshot taken after an action of a step, as an image, to check the screen instead of the reasoning text. By default, the last action of the step that has one.

runId, stepIndex. Optional: actionIndex.

get_test_case_history

Read

The last runs of one test case, newest first, with their status and environment.

testCaseId. Optional: since, environmentId, limit (default 20, maximum 100).

Environments

Tool

Access

What it does

Parameters

get_environments_by_project_id

Read

List a project's environments.

projectId

get_environment_by_id

Read

Fetch one environment with its variables and credentials. Secret values are masked.

environmentId

create_environment

Write

Create an environment and point each app at a URL.

projectId, name. Optional: description, apps (an appId plus url per app).

update_environment

Write

Rename an environment or change its per-app URLs.

environmentId, name. Optional: description, apps.

clone_environment

Write

Copy an environment with all its variables and credentials. Defaults to "(Clone)".

environmentId. Optional: name.

Environment variables and credentials

Tool

Access

What it does

Parameters

create_environment_variable

Write

Add one variable or secret credential.

environmentId, projectId, name, value, isSecret. Optional: description. Names are uppercase A to Z, 0 to 9 and underscores, max 30 characters.

create_environment_variables

Write

Add many variables in one call.

A list of the above.

Authenticated Profiles

Tool

Access

What it does

Parameters

list_authenticated_profiles

Read

List the saved Authenticated Profiles of a project. A test can restore a profile instead of running login steps. A profile whose expiresAt has passed must be recorded again.

projectId

Personas

Tool

Access

What it does

Parameters

get_all_personas

Read

List the personas available for execution.

None

Did this answer your question?