MCP App Store
by KeenEthics
QA Wolf icon

QA Wolf MCP Server for Claude and ChatGPT

By QA Wolf Inc.

Overview

Connect your coding agent to QA Wolf to create, run, and troubleshoot end-to-end tests. Explore your app in a cloud browser, review a proposed test outline, and approve QA Wolf to implement and validate it. Sign in with your QA Wolf account and follow progress through live session links.

How to connect

  1. Claude: Open the QA Wolf listing in the Claude connectors directory and choose Connect. Claude listing QA Wolf exposes 49 tools here.
  2. ChatGPT: Open the QA Wolf listing in the ChatGPT app directory and choose Connect. ChatGPT listing QA Wolf exposes 111 tools here.

Tools

activity_wait

ChatGPT
Wait on many QA Wolf AI (Tester) sessions and runs in one call. Pass every Tester session and run you are waiting on. The call returns as soon as any session replies or changes status or any run finishes, and reports every item with news. Send back open as the next call's input, and call again as soon as it returns. Pass waitSeconds above 45 only when your instructions say your client keeps the call open while progress arrives.

agent_get

ChatGPT
Monitor a QA Wolf AI session by reading its status and replies. After agent.send, share the returned session URL before monitoring. Pass waitSeconds 45 so the server holds the check open and answers the moment something happens. Check again as soon as it returns; never sleep between checks or loop without waitSeconds. Pass the nextCursor from one response as the cursor on the next check; it then reads only what is new, and only for the session that minted it. Continue monitoring silently when the status is unchanged and no replies come back; do not narrate waiting, announce the next check, or ask whether to keep monitoring. Report only substantive new progress, questions, blockers, or the final outcome. A status of "waiting-for-you" means the last reply is a question the work is blocked on, and answering it with agent.send is what unblocks it. Surface an explicit request for user input even if the status still says "working". Include the session URL when reporting a blocker or final outcome. On "completed", stop status checks and verify the requested result before claiming success. For new flows, validation, publication in the target environment, and readiness are separate checks; a Git push or final reply does not prove the flow is active. If every requested result is verified but status remains "working", report the mismatch and stop monitoring. Stop on "failed" or "cancelled" and report any confirmed partial result.

agent_send

ChatGPT
Start or continue work with the QA Wolf AI and return a live session URL to share with the user. Use it to cover a user journey, investigate a failing run, or fix a broken flow. This is the one verb that starts work from nothing: every other write acts on a flow, run or issue that already exists. Returns sessionId, status, and url as soon as the request is accepted; work can take minutes to tens of minutes. After each send, put the exact returned url in a normal user-visible assistant message, and repeat it in the last message of the turn: only a turn's trailing message is shown expanded. Tool output and internal reasoning do not count as sharing the link. Do not run a timer or monitoring call alongside this send. Acceptance does not mean the work is complete. Then monitor the session with agent.get, reporting new progress, blockers, and the final outcome rather than unchanged status. Send here again to answer a question or add context to the same session.

attempt_get

ChatGPT
Get a finished run attempt: its run and flow status, its place among that flow's attempts in the run, and signed links to its logs, trace, and video when they were captured.

build_create

ChatGPT
Create a build from a file uploaded with build.requestUpload. Reads the app ID, name, version and build number from the file before it answers, or refuses a file that isn't an app. Calling it again with the same buildId answers the build the first call created, unchanged: a repeated call's branch, commitSha and environment are ignored.

build_delete

ChatGPT
Delete a build. Runs never install a deleted build. Deleting an environment's latest build makes the previous one its latest. Runs install that build only if its file is still kept: only the latest builds of static environments are kept past the bucket's expiry, 180 days after upload on QA Wolf's cloud. Answers the build with deletedAt set. Deleting a deleted build answers it unchanged.

build_get

ChatGPT
Get a build: its app, platform, version and environment. Deleted builds are returned too.

build_list

ChatGPT
List the workspace's undeleted builds, newest first. With latestOnly, list what the environment's next run installs.

build_requestDownload

ChatGPT
Get a URL to download a build's file. Deleted builds can be downloaded while their file exists. Answers 404 when the file has expired.

build_requestUpload

ChatGPT
Get a URL to upload an app build: an .apk, .aab or .ipa, up to 500 MB. PUT the file there with uploadHeaders, then call build.create with the buildId.

codeHostIntegration_find

ChatGPT
List the workspace's code host integrations (GitHub or GitLab). A connected integration lets QA Wolf read repositories and receive the code host's webhooks. codeHostIntegration.listRepositories pages the repositories each integration covers.

codeHostIntegration_findForRepository

ChatGPT
Check whether QA Wolf can read a repository through the workspace's code host integrations, and what it can access: the repository's code and its pull requests. A CI job calls this before reporting a deployment to decide which changes to include. Not covered means QA Wolf only knows the changes the report includes; settingsUrl is where a workspace admin connects the code host.

codeHostIntegration_listRepositories

ChatGPT
List the repositories the workspace's code host integrations cover, alphabetical by full name. The list reflects the last sync from the code host; codeHostIntegration.find lists the integrations themselves.

deployment_find

ChatGPT
List the deployments QA Wolf has received for the workspace, newest first. A deployment arrives through a code host integration's webhook or through deployment.reportStatus, and is what a deployment trigger evaluates against. deployment.listTriggerEvaluations reports each trigger's verdict for one deployment.

deployment_listTriggerEvaluations

ChatGPT
List the per-trigger verdicts recorded when a deployment was evaluated against the workspace's triggers. Each verdict says whether the trigger matched, did not match with the reason per condition, or was never considered and why. Verdicts are a snapshot from evaluation time: editing or deleting a trigger later does not change them.

deployment_reportStatus

ChatGPT
Report a deployment lifecycle status. A deployment's first success report evaluates global triggers asynchronously, and it is the only report that does: a trigger added or unpaused later does not make an already reported deployment evaluate again. Report that deployment under a new providerDeploymentId to evaluate it against the current triggers. The response contains the deployment only, not the resulting runs.

email_find

ChatGPT
List the workspace's inbox, or its sent mail, newest first. Read a message body with email.get.

email_get

ChatGPT
Read one email of the workspace, with its plain text and HTML bodies. Use it to pull a sign-in code or a verification link out of a message.

email_getAttachment

ChatGPT
Read one attachment of a workspace email as base64 content, by file name or by position. email.get lists both.

email_listAddresses

ChatGPT
List the workspace's inbox addresses, alphabetical. A flow can sign up with a plus-suffixed form of any of them, and email.find reads what arrives.

email_registerAddress

ChatGPT
Register an inbox address for the workspace. Registering an address the workspace already has changes nothing. A refusal names the domains the workspace can use.

email_send

ChatGPT
Send an email from one of the workspace's inbox addresses, for example to exercise a flow that reacts to incoming mail. Returns the sent email; read it back with email.get.

environment_create

ChatGPT
Create an environment on the caller's team and return it in the environment.get shape. This can also create a branch on the team's connected Git provider.

environment_deleteVariable

ChatGPT
Remove one environment variable by name. Succeeds whether or not the variable existed.

environment_find

ChatGPT
List the team's environments, newest first.

environment_get

ChatGPT
Read a single environment's name, kind, standing run health, flow-code branch and reconciliation state, run concurrency limit, and termination state. If flowCodeBranch exists, use its syncStatus for Git reconciliation and read lastSyncedCommitHash only when syncStatus is reconciled.

environment_getVariable

ChatGPT
Read the values of named environment variables in one call. Values are secrets. Names that do not exist go to missingNames and do not fail the call.

environment_listVariableNames

ChatGPT
Use this to answer which QA Wolf environment variables are available to test code. Returns names only; values never leave the server.

environment_promote

ChatGPT
Declare a preview environment's work accepted, typically when its pull request merges: its flow changes are promoted into its base environment and the environment is terminated once they land. Returns the promotion in pending; poll promotion.get to learn how it settled. Calling again while a promotion is open returns that promotion. A conflict settles the promotion conflicted and cancels the environment's in-progress runs; promotion.get names the conflicting paths. Use environment.terminate instead when the work is abandoned.

environment_setVariable

ChatGPT
Create or replace an environment variable. If the user asks to create one for "my email" without naming it, use DEFAULT_EMAIL. The value is never returned.

environment_terminate

ChatGPT
Terminate a preview environment, typically when its pull request closes without merging: its flow changes are not promoted into its base. Returns the environment, now terminated. A terminated environment cannot be terminated again. Use environment.promote instead when the work is accepted.

environment_update

ChatGPT
Update an environment owned by the caller's team and return it in the environment.get shape. Omitted fields remain unchanged. At least one of name or runConcurrencyLimit is required.

file_requestDownload

ChatGPT
Get a URL for reading a file out of the caller's team storage. Answers 404 when nothing is stored at that path.

file_requestUpload

ChatGPT
Get a URL to put a file into team storage: a spreadsheet of journeys, anything too large to paste. PUT with the returned contentType, then name the returned filePath in filePaths.

flow_addTag

ChatGPT
Assign an existing tag to the selected flows. Create tags with tag.create. Flows that already carry the tag are reported in skippedFlows.

flow_getHistory

ChatGPT
List one flow's results across runs in an environment, newest first, with status, attempts, duration, and diagnosis. Use it to spot flakiness, failure streaks, and duration trends.

flow_list

ChatGPT
List the flows of an environment at its latest reconciled commit, or, when an AI task is given, the flows on that task's branch. An empty flows array is not on its own evidence that QA Wolf lost a flow. Read emptyReason, report what it says, and claim no more than it establishes.

flow_removeTag

ChatGPT
Remove a tag from the selected flows. Succeeds whether or not each flow carried the tag; the flows that did not are reported in skippedFlows.

flow_update

ChatGPT
Move a flow between draft and active readiness. The other statuses shown in the app are derived and cannot be set.

investigation_get

ChatGPT
Read what the QA Wolf AI investigating a run has concluded: one finding per cause of its failures, with the question waiting for a person and any answer or dispute the investigation hasn't acted on yet. A run that automatic investigation does not cover answers status "not-enabled".

investigation_question_answer

ChatGPT
Answer the questions the QA Wolf AI investigating a failed run is waiting on, as the signed-in person. List the questions with investigation.question.find, and answer only with the person's approval. The first answer to a question wins, and the investigation acts on it. Only a person's own sign-in or user API key can answer; a team or organization key cannot.

investigation_question_find

ChatGPT
List the questions the QA Wolf AI investigating failed runs is waiting for a person to answer, across a workspace's runs. Each question names its run, flows, details and offered answers. To answer them, call investigation.question.answer with the run and the person's answers.

issue_addFlows

ChatGPT
Add flows to a coverage request owned by the caller's team. Flows already covered stay covered. Bug and maintenance reports link to flows through the runs that reproduce them; use run.diagnose to record one.

issue_attachFiles

ChatGPT
Attach files from the team's storage to a bug report owned by the caller's team, for example the files a bug was reproduced with. Files already attached stay attached. Upload a file with file.requestUpload first.

issue_comment

ChatGPT
Comment on a bug report, blocker (issue type maintenance), or coverage request owned by the caller's team. The comment changes nothing else on the issue, such as its status or priority.

issue_create

ChatGPT
Create a bug report, maintenance report, or coverage request issue for the caller's team. A user can create a maintenance report only when they can see the workspace's maintenance reports; a team or organization API key credits it to the workspace's automation user. Link its failed flows afterwards with run.diagnose.

issue_decideSuggestion

ChatGPT
Decide a coverage request or bug report that QA Wolf suggested after exploring an environment. Add turns on the suggested flow in the environment. Report files a potential bug. Skip declines the suggestion. Undo takes back a decision before the flow joins. The issue's status follows the decision, so change the status only through this tool while the suggestion is undecided.

issue_find

ChatGPT
List the team's bug reports, maintenance reports, or coverage requests, newest first.

issue_get

ChatGPT
Get an issue by id.

issue_listActivity

ChatGPT
List an issue's activity, oldest first: creation, status and priority changes, and the flows that started or stopped reproducing it. Use it to measure time to fix or to find issues that came back.

issue_listComments

ChatGPT
List the comments on a bug report, blocker (issue type maintenance), or coverage request, oldest first, with who wrote each one.

issue_markFixReadyForReview

ChatGPT
Mark the draft pull request of the bug's fix ready for review, so the code host asks its code owners. The pull request stays a draft unless a loop's instructions ask for this. Only a QA Wolf session that a loop started can call it. Refused as no-fix-pull-request when the bug's fix has no open pull request.

issue_removeFlows

ChatGPT
Remove flows from a coverage request or remove bug/maintenance reproductions in an environment. Unlinked flows are left alone. Reports auto-close when no active reproductions remain and auto-close is enabled. For reports, omit environmentId only when each selected flow is linked in at most one environment.

issue_requestFix

ChatGPT
Hand a bug report to the team's own coding agent, which fixes it in the team's CI. Only a QA Wolf session that a loop started can call it, when the loop's instructions use /coding-agent. A bug has one open fix: a second request returns that fix.

issue_retryFix

ChatGPT
Ask the team's coding agent for one more try at a bug that the check of its fix still shows. QA Wolf sends the same agent the failure of the check's run, as a new fix on the same branch, and checks again. Only a QA Wolf session that a loop started can call it, when the loop's instructions ask for it. A bug gets one extra try at most. Refused as bug-not-still-showing when the check of the bug's fix does not show the bug, and as no-tries-left after the extra try.

issue_update

ChatGPT
Update an issue owned by the caller's team. Omitted fields remain unchanged. At least one of description, name, priority or status is required.

legacyTrigger_find

ChatGPT
List a workspace's legacy per-environment triggers; trigger.find lists the current ones. For migration only; it goes when legacy triggers go. adaptiveFlowSelection becomes a generativeSuite action, which takes no tags.

legacyTrigger_pause

ChatGPT
Pause one legacy trigger, the older per-environment kind, so it stops firing; use trigger.pause for a current one. Its copies in pull request environments pause too. For migration only; it goes when legacy triggers go.

legacyTrigger_resume

ChatGPT
Resume one legacy trigger, the older per-environment kind; use trigger.resume for a current one. Its copies in pull request environments resume too, and a scheduled one jumps to the next upcoming slot even if it was not paused. For migration only; it goes when legacy triggers go.

mcpServer_connect

ChatGPT
Start connecting an MCP server from the workspace's catalog, such as Linear, so Tester can use its tools. Returns a sign-in link to show the user.

narratedRecording_get

ChatGPT
Get a narrated recording's voiced video: its video page URL, its export status, and, once the export is ready, its duration and signed links to the mp4 with the voiceover baked in and its subtitles. Signed links stop working at video.expiresAt; the video page URL stays stable. While exportStatus is pending, check again later.

promotion_find

ChatGPT
List environment promotions, newest first. A promotion moves the flow changes made in a head environment into its base environment. Filter by status conflicted to find the promotions whose conflicts still need a resolution.

promotion_get

ChatGPT
Read one environment promotion: its status, the conflicting paths and commit hashes while it is conflicted or resolving, the commit it landed when promoted, and the AI task resolving it. Poll it after environment.promote to learn how the promotion settled. It does not serve file contents; a resolver reads both branches from the flow code repository.

promotion_resolve

ChatGPT
Submit the resolution of a conflicted promotion: the full resolved contents of every conflicting path to keep, and every conflicting path to delete. Together they must cover exactly the paths in the promotion's conflict, and both commit hashes must match the conflict's; if a branch moved since, the call is refused as stale-conflict and promotion.get has the current conflict. Returns the promotion in resolving; poll promotion.get until it is promoted, or conflicted again if the branches moved while it applied.

propose_outline

ChatGPT
Draw an Arrange-Act-Assert flow outline in a card, and answer with the same outline as text. Put every outline in the batch in a single call. This starts no work, creates nothing and bills nothing. Then send that same outline as a normal message too, because most clients draw no card and the message is the only place their user reads it. Ask which outlines to create after that, and once the user approves, send the same outline text to QA Wolf with agent_send.

run_acceptScreenshotBaseline

ChatGPT
Make the new screenshot of a failed screenshot comparison the baseline, so later runs compare with it. Use it only when the difference is an intended change of the app, not a bug, noise, or a page that was not ready. Every later run of every environment of the workspace compares with the new baseline. Refuses without changing anything when the baseline changed since the run.

run_create

ChatGPT
Create a run for the selected flows and/or tags in an environment. At least one flow or tag is required, so send flowIds, tagNames, or both. Send at most one of aiTaskId and chatSessionId. This call has no idempotency key, so a lost response cannot be safely resubmitted: read run.find before sending it again.

run_diagnose

ChatGPT
Link failed flows in a run to an existing bug or maintenance report owned by the caller's team, recording them as its reproductions. It does not explain why a run failed; read investigation.get or attempt.get for that. The issue's type selects the diagnosis. Each flow must have failed in the run. A flow that is already diagnosed in the run moves to this issue, and may stay linked to other open reports. The diagnosis appears on the run, and the reproduction appears under the issue's reproductions. Coverage requests cannot be diagnosed; use issue.addFlows to cover flows instead.

run_find

ChatGPT
List an environment's recent runs, newest first, including failed-attempt discovery metadata.

run_get

ChatGPT
Get a run's status, per-flow results, links, and how many of its bugs are blocking. Read a run with flowStatuses: ["failed"] first. A flow that passed on a retry reads passed, so when the flow you want is missing, add "passed" to the filter.

run_listScreenshotComparisons

ChatGPT
List the screenshot comparisons that failed a run's flows, with the expected, actual and diff images of each. Only comparisons that caused a failure are listed: a screenshot that differs within tolerance, or that differs on a line that did not fail, is left out. Each comparison tells whether its baseline changed since the run and whether the baseline still passes in other environments. The image URLs expire after a few minutes; fetch the preview URLs to look at the images. The first call for an image makes its missing JPEG preview and keeps it in team storage, so later calls reuse it.

run_optOutOfInvestigation

ChatGPT
Mark failed flows in a run as "do not investigate" (DNI): the failures need no bug or maintenance report. Each flow must have failed in the run and still be under investigation. A flow that already has a diagnosis cannot be opted out. A team or organization key records the opt-out as the workspace's automation user. To link a failure to a report instead, use run.diagnose.

run_reattempt

ChatGPT
Request new attempts for a run's flows. A flow is eligible once its result is failed or canceled and QA Wolf's automatic retries have finished. A flow that kept failing in later runs is attempted in the latest of them, because that is the run carrying the failure, so read reattemptedInRunIds for where each attempt landed. A fully investigated run no longer accepts reattempts. Attempts run with the flow code the environment has synced when the request arrives: after publishing a fix, pass afterCommit so the request waits for that commit instead of rerunning older code. Poll run.get for results, or pass aiTaskId or chatSessionId to be notified when they settle.

run_restoreScreenshotBaseline

ChatGPT
Undo a screenshot baseline change that accepted a failed comparison of the run: put back the baseline it replaced, so later runs compare with the old baseline again. Use it when a person disputes a baseline the AI accepted. Refuses without changing anything when the baseline changed again since.

run_stop

ChatGPT
Stop a run, including its queued flows and automatic retries. Stopping is asynchronous and can update run-status messages and commit statuses in connected integrations. Repeated requests are safe, and finished runs keep their results. A run that is still being created returns not found; retry once run.get returns the run. If run.get returns a different runId, use that ID. Poll run.get for results.

run_view

ChatGPT
Open QA Wolf for the user to explore workspaces, bugs, maintenance reports, runs, runners, flows, and the other MCP capabilities. Users can perform existing actions and explicitly add items to chat context. Opens from the sidebar, conversation panel, and QA Wolf links. Takes no arguments. To read data yourself, use the relevant MCP tools.

runner_evaluateSnippet

ChatGPT
Evaluate a snippet against whatever the runner's browser is showing right now. Answers whether the snippet ran, and its error if it threw — not the value it evaluated to, so read anything you want back out of the runner's journal (a snippet's console.log lands in the console stream). A snippet starts no run and leaves no run-scoped journal entries of its own. runner-cannot-evaluate-snippets on a runner with nothing attached to serve one — which will never clear, so check that the runner you launched is one that runs a browser or app before retrying. runner-unreachable if the runner could not be reached: it may still be starting, it may have terminated after inactivity, or it may be too busy to answer. Neither is proof the snippet did not run: a snippet that outlives the answer window is still executing when you read this, so do not blindly resubmit one that mutates what the page is looking at.

runner_get

ChatGPT
Report whether a runner is running under this id on the caller's team. This is a lookup: it never starts a runner, and it does not reset a runner's inactivity clock the way reading its journal does. An id that is not running is a success with running: false rather than an error, so this is how to check a runner is still there before addressing it. A running runner also reports the url to watch it on, so an id alone is enough to recover the link to give the user.

runner_highlightSelector

ChatGPT
Highlight the elements a selector matches on an interactive runner's live page, and answer how many it matched. The highlight stays until it is replaced or cleared, so it is visible in the next runner.takeScreenshot — which is the point, since a caller cannot see the runner's screen otherwise. Send an empty selector to clear, which answers cleared. status tells a selector that matched nothing (empty) from one the page could not parse (invalid), so a caller can tell a bad locator from a locator pointing at nothing. no-answer if the page did not answer in time: the highlight runs inside the page, so a page that is gone or mid-navigation does not answer slowly, it does not answer at all. runner-cannot-highlight-selectors on a runner with no browser to draw on. runner-unreachable if the runner could not be reached: it may still be starting, or it may have terminated after inactivity.

runner_importPackage

ChatGPT
Install a package into an interactive runner's live run and import it, so a snippet or a selection can use it without a full run to reinstall dependencies. An install-failed failure carries npm's reason in errorMessage. runner-unreachable if the runner could not be reached: it may still be starting, it may have terminated after inactivity, or it may be too busy to answer. The install may still have landed, but installing the same version again does nothing, so retrying is safe.

runner_inspect

ChatGPT
Inspect one thing on an interactive runner: an element's HTML, the page's HTML simplified for a model, or a top-level variable's value as JSON. This refreshes the inactivity timer and can extend the runner's billed lifetime. nothing-to-inspect means the runner had nothing to answer with: no live page, no element matching the selector, or no variable under that name. A runner cannot tell those apart, so it reports the one reason and puts whatever it did say in errorMessage. A mobile or Windows runner answers variable; for element-html or page-html it answers runner-is-not-a-browser — call runner.inspectMobile for its equivalent surface instead. runner-unreachable if the runner could not be reached: it may still be starting, it may have terminated after inactivity, or it may be too busy to answer. Nothing was changed, so retrying is safe; if it persists, launch the runner again.

runner_inspectMobile

ChatGPT
Inspect one thing on a mobile or Windows interactive runner: the Appium session's status, the contexts available (WebViews on mobile, top-level windows on Windows), the current context's page source, or the elements at a point, carrying some text, or matching a selector. This does not refresh the runner's inactivity timer. Mobile and Windows only — a browser runner answers runner-is-not-mobile; call runner.inspect for a browser's equivalent surface instead. what: "session" always answers with the session's own status rather than screen-needs-a-run, since that is the question it exists to answer; the other three request kinds need a live session first and answer the same screen-needs-a-run or screen-not-ready outcomes runner.performAction and runner.takeScreenshot use, instead of reading anything when there is none. screen-needs-a-run means no Appium session has started on this runner yet — call runner.runFlow with a flow that opens one, then inspect again. screen-not-ready means the runner's Appium session exists but did not answer this instant, or more than one is somehow live — retry once; if it persists, relaunch the runner. runner-is-not-mobile means there is nothing here to inspect, and retrying will never help — launch an android, ios or windows runner instead. invalid-selector (elements by selector only) means the selector itself could not be parsed under the given strategy — distinct from a selector that parsed fine but matched nothing, which answers matches: [] instead. runner-unreachable if the runner could not be reached: it may still be starting, it may have terminated after inactivity, or it may be too busy to answer. Nothing was changed, so retrying is safe; if it persists, launch the runner again.

runner_launch

ChatGPT
Launch an interactive runner on the caller's team under an id the caller chooses. Send initialUrl so its browser comes up on the page the work starts from, which saves a separate navigation and the browser's cold start. Launching the same id again returns the runner already running rather than starting a second one, and the same id with a different runnerName is refused. A runner is not permanent: it terminates on its own after a period of inactivity, and launching the same id after that starts and bills a new runner, so read alreadyRunning to tell which happened. A success means the runner is answering: the call waits for the pod it started to come up, so the very next call to the runner reaches it. If the pod does not come up in time the call fails instead, leaving the runner running, and launching the same id again attaches to it.

runner_list

ChatGPT
List the runners running on the caller's team right now. A listed runner may be driven by another machine or session, so sending it work can collide with theirs. A runner that was terminated, or that terminated on its own after inactivity, is not listed, because the run system keeps no record of a runner once it is gone. This is a lookup: it never starts a runner, and it does not reset a runner's inactivity clock the way reading its journal does. It is how to find a runner launched from another machine or in an earlier session, when the id it was launched under is no longer at hand.

runner_performAction

ChatGPT
Perform one raw browser action on an interactive runner: click, double_click, move, drag, scroll, keypress, type, or navigate. One action per request, and the runner serves one at a time. A success means the action took effect. action-failed, with a reason, if it reached the runner and did not take effect. On a runner image with a browser, the first action on a runner that has never run anything starts its browser and waits for it, so it can take up to a minute to answer — no runner.runFlow is needed before acting; if the browser is still starting when the wait runs out, the answer is screen-not-ready and retrying converges. On a mobile runner, the same screen-needs-a-run and screen-not-ready outcomes mean no Appium session has started yet, or it did not answer this instant; runner.runFlow is what starts one, same as a browser. screen-needs-a-run if the browser could not be started that way — usually a runner whose runs all finished without starting its desktop; call runner.runFlow with a flow that opens a browser. screen-not-ready if the runner has a screen that cannot serve this instant. Retry in a second or two: the desktop restarts when a run changes the display size, and it serves one see-or-act request at a time, so a screenshot or action already in flight is the usual reason. Do not submit a run to clear this — a run may restart the display and discard what is on it. runner-has-no-screen if this runner is not one that runs a browser on a virtual desktop. Nothing about it can be seen or driven, and retrying will never help — launch a playwright runner instead. runner-unreachable if the runner could not be reached: it may still be starting, it may have terminated after inactivity, or it may have stopped answering mid-action. This does not mean the action was not performed — take a screenshot before repeating it. Send withScreenshot: true for the screen the action led to in the same answer, rather than a second round trip through runner_takeScreenshot.

runner_performActions

ChatGPT
Perform a sequence of raw browser actions on an interactive runner in one request, one after another, and answer with what happened to each. A success means every action took effect. A failure names the first action that did not (failedIndex, failureReason), the last one that did (lastCompletedIndex), and whether actions were left unperformed (stoppedEarly); results has one entry per action reached, in order. Read effect on an entry before sending its action again: performed means the runner did it; not-performed means the runner answered that it did not take effect, though a navigate that timed out may still be loading; unknown means the runner stopped answering with the action in flight (runner-unreachable), or its screen went quiet mid-action (action-unconfirmed), and it may have taken effect — a click that submitted a form, say — so take a screenshot instead of repeating it. A sequence gets about a minute to start its actions; an action it did not reach in time is out-of-time, nothing after it was attempted, and the rest can go in a new request. Everything runner.performAction says about a runner that has never run anything, a mobile runner, and the screen-needs-a-run, screen-not-ready, runner-has-no-screen and action-not-supported-on-mobile reasons holds for each action here.

runner_promoteSnapshot

ChatGPT
Accept a run's screenshot as the new baseline for an image diff, on the runner that produced it. The two paths are the ones the diff reported, which reach a caller as an image-diff-artifact entry on the runner's run-events journal stream. snapshot-not-found if the run wrote no screenshot at that path, which usually means the paths were not taken from a diff this run produced; nothing is changed, so correcting the path and repeating is safe. runner-cannot-promote-snapshots on a runner that stores no screenshots. runner-unreachable if the runner could not be reached: it may still be starting, or it may have terminated after inactivity. A promotion that lands is not undone by promoting again, so a retry after an unreachable answer is safe.

runner_readJournal

ChatGPT
Read a window of one of an interactive runner's journal streams — the newest few, everything after a cursor, or everything belonging to one run. This is how a flow run's outcome and output are followed: run-status settles it, run-logs and run-events carry what it produced, and recorder carries the browser actions the runner recorded. A read counts as activity, so working through history does not get the runner reaped underneath you. runner-unreachable if the runner could not be reached: it may still be starting, it may have terminated after inactivity, or it may be too busy to answer. Nothing was changed, so retrying is safe; if it persists, launch the runner again.

runner_record

ChatGPT
Control video recording on a Playwright runner. Start requires a caller-generated UUID; stop targets that UUID so retries cannot stop a later recording. Manual capture spans runs and suppresses automatic capture. Auto settings affect subsequent full runs. The URL opens recording history after the runner terminates. Videos may be unavailable after abrupt pod loss.

runner_recordings

ChatGPT
Read a runner's finalized recordings from workspace storage, including after the runner terminates. Omit recordingId to list recordings in storage-key order; follow nextPageToken for more. An empty list for a specific recording means it is not published or does not exist. Video URLs expire; platform URLs remain stable.

runner_runFlow

ChatGPT
Run a flow on an interactive runner. When selection is given the lines run against the browser as it stands, so nothing is re-navigated and nothing is signed in again; without it the whole entry point runs from a fresh browser. A runner with no live browser starts one before a selection and says so with bootstrappedRunner, which means those lines ran against a fresh page rather than the one an earlier run left. A selection produces no runStarted event, so follow it by its run id on run-status like any other run. Send unchangedFiles to ship only what changed since an earlier run on this runner; a needs-full-sync failure names the paths it does not hold, and the way to recover is the same run again with every file in files. Answers as soon as the run is accepted, with the id to follow it by — nothing waits for the run to finish. Which browser or device the run needs is read from the flow file's own execution target, so it is not supplied here; when it does not match what the runner is, the call fails with runner-target-mismatch rather than failing partway through the run. runner-unreachable means the answer did not arrive, which is NOT the same as the run not having started: the runner may have accepted it and been too slow to say so, and resubmitting would start a second run that is billed and journalled alongside the first. Read the runner's run-status journal stream before resubmitting, and use the newest run id there if one appeared. files must include the entry point and the run's package.json. Send env or environmentId, not both. A path may be sent in files or referenced in unchangedFiles, not both. This call has no idempotency key, so a lost response cannot be safely resubmitted: read the runner's run-status journal stream before sending it again.

runner_stopRun

ChatGPT
Stop what a runner is currently executing, leaving the runner up and its browser on whatever page the run reached. Succeeds whether or not anything was running, and wasRunning says which. This is the counterpart of runner.terminate, which ends the runner itself. The run stops where it is, so the journal's run-status settles it as stopped rather than passed or failed. runner-unreachable if the runner could not be reached: it may still be starting, it may have terminated after inactivity, or it may be too busy to answer. The stop may still have landed, but stopping a runner that is already idle does nothing, so retrying is safe.

runner_takeScreenshot

ChatGPT
Take one screenshot of an interactive runner's screen. On a runner with a browser the image is the whole virtual desktop, browser window and all; this refreshes the inactivity timer and can extend the runner's billed lifetime. On a mobile runner it is the device's own screen, and on Windows the VM's whole desktop, re-encoded to JPEG on the pod so this contract reads one image format regardless of runner family. screen-needs-a-run if the runner's virtual desktop has never started. Waiting will not change this and retrying is pointless — call runner.runFlow with a flow that opens a browser, then ask for the screen again. On a Windows runner any flow run claims the VM and brings up its screen. On a runner image with a browser that has never run anything, any runner.performAction also starts the browser itself. Evaluating a snippet does not start the desktop. screen-not-ready if the runner has a screen that cannot serve this instant. Retry in a second or two: the desktop restarts when a run changes the display size, and it serves one see-or-act request at a time, so a screenshot or action already in flight is the usual reason. Do not submit a run to clear this — a run may restart the display and discard what is on it. runner-has-no-screen if this runner has no browser desktop to see or drive, and retrying will never help. Launch a playwright runner. runner-unreachable if the runner could not be reached: it may still be starting, it may have terminated after inactivity, or it may be too busy to answer. Nothing was changed, so retrying is safe; if it persists, launch the runner again. To see the screen an action led to, send withScreenshot: true on runner_performAction instead, which answers with the frame; call this on its own to look at the screen without acting on it.

runner_terminate

ChatGPT
End an interactive runner on the caller's team, and the pod it runs on with it. Terminating a runner that is not running succeeds and reports wasRunning: false. This ends the runner: use runner.stopRun to stop what a runner is currently executing while leaving the runner up.

skill_get

ChatGPT
Read a QA Wolf skill: the instructions a coding agent follows for one kind of QA Wolf work. The reply carries the skill's SKILL.md and every file under its references directory, so nothing else has to be fetched for it. Read the skill before starting the work it covers and follow it. A link in the reply of the form ../<skill>/SKILL.md names another skill, which this call reads by that name.

skill_list

ChatGPT
List the QA Wolf skills, each with its name and the description that says when to use it. A skill is the instructions a coding agent follows for one kind of QA Wolf work, such as onboarding an application, creating a flow, repairing a failing flow, or setting up triggers. Call this before any QA Wolf work, pick the skill whose description matches the request, and read it with skill.get.

suggestion_decide

ChatGPT
Decide a coverage suggestion QA Wolf made while exploring an environment, as the signed-in person. Add turns on the suggested flow in the environment. Report files the potential bug. Skip declines the suggestion and needs a skipReason. Undo takes back a decision. Activate retries a held suggestion. Only the decisions suggestion.get lists take effect. Use this, not issue.update, to accept or skip a suggestion.

suggestion_find

ChatGPT
List the coverage QA Wolf suggested for a workspace, newest first: flows it wrote for missing coverage and potential bugs it found while exploring an environment. Use this, not issue.find, for QA Wolf's suggestions, such as the ones a new suggestions notification points to. Filter by state waiting to find the ones that need a decision.

suggestion_get

ChatGPT
Read one coverage suggestion QA Wolf made while exploring an environment: its state, the flow it proposes, the bug report or coverage request behind it, and the decisions it takes now. Use this, not issue.get, for a suggestion id. A suggestion QA Wolf withdrew reads as not found.

tag_create

ChatGPT
Create a tag on the caller's team. Tags select flows in run.create.

tag_list

ChatGPT
List the team's tags, alphabetical by name. Tag names select flows in run.create.

trigger_create

ChatGPT
Create a trigger. A schedule trigger runs a named set of flows on a cadence; a deployment trigger runs when a matching deployment is reported.

trigger_delete

ChatGPT
Delete a trigger permanently. The runs it already created are kept. To stop a trigger without losing it, use trigger.pause.

trigger_find

ChatGPT
List the team's triggers, newest first.

trigger_get

ChatGPT
Get one trigger by id.

trigger_listFirings

ChatGPT
List when a trigger fired and what each firing started: a run, a QA Wolf session, or nothing and why.

trigger_pause

ChatGPT
Pause a trigger so it stops firing. Pausing is idempotent. A team whose triggers are all paused reports no trigger activity at all.

trigger_resume

ChatGPT
Resume a paused trigger. A schedule trigger restarts from the next upcoming slot: the slots it missed while paused do not run.

trigger_runNow

ChatGPT
Fire a trigger once by hand. It works on a paused trigger too, and does the same as any other firing, except that a run it starts is not linked to the deployment or its pull request. Follow the firing with trigger.listFirings.

trigger_update

ChatGPT
Replace a trigger's configuration. Every field is written, so read the trigger first and send its configuration back with your changes applied. Pausing is separate: use trigger.pause and trigger.resume.

usage_breakdown

ChatGPT
Show where credits went over a range of UTC days, grouped by time, workspace, environment, platform, source, trigger or user. Use it to find what uses the most credits, or to build a usage dashboard. Credits only, no dollar amounts.

whoami

ChatGPT
Report the workspace, team, organization or user authenticated on this QA Wolf connection. The reply lists authorized workspaces with their organization and app slug. Match a link's slug against this list to obtain its workspaceId. Unbound connections require workspaceId on other tools. Use a workspace explicitly selected by the user or provided in current QA Wolf app context when it appears in this reply. Do not ask again just because several workspaces are listed. If the provided workspace is absent, check that the model and app use the same QA Wolf connection and deployment. Never substitute a different workspace. When no workspace is selected and several are available, ask the user which one: offer a few choices and let them name another. Make that question the last thing in your turn.

agent_get

Claude

agent_send

Claude

automate

Claude

email_find

Claude

email_get

Claude

email_getAttachment

Claude

email_listAddresses

Claude

email_send

Claude

environment_create

Claude

environment_deleteVariable

Claude

environment_find

Claude

environment_get

Claude

environment_getVariable

Claude

environment_listVariableNames

Claude

environment_setVariable

Claude

environment_update

Claude

flow_addTag

Claude

flow_list

Claude

flow_removeTag

Claude

flow_update

Claude

issue_addFlows

Claude

issue_create

Claude

issue_find

Claude

issue_get

Claude

issue_removeFlows

Claude

issue_update

Claude

run_create

Claude

run_diagnose

Claude

run_find

Claude

run_get

Claude

run_reattempt

Claude

runner_evaluateSnippet

Claude

runner_get

Claude

runner_highlightSelector

Claude

runner_importPackage

Claude

runner_inspect

Claude

runner_inspectMobile

Claude

runner_launch

Claude

runner_list

Claude

runner_performAction

Claude

runner_promoteSnapshot

Claude

runner_readJournal

Claude

runner_runFlow

Claude

runner_stopRun

Claude

runner_takeScreenshot

Claude

runner_terminate

Claude

tag_create

Claude

tag_list

Claude

whoami

Claude

Similar Apps

Listed on MCP App Store

Listed on MCP App Store

Add this badge to your site or README. It links to this page.

HTML

<a href="https://mcp-app-store.keenethics-labs.com/app/qawolf"><img src="https://mcp-app-store.keenethics-labs.com/badges/listed-on-mcp-app-store.svg" alt="Listed on MCP App Store" width="220" height="44"></a>

Markdown

[![Listed on MCP App Store](https://mcp-app-store.keenethics-labs.com/badges/listed-on-mcp-app-store.svg)](https://mcp-app-store.keenethics-labs.com/app/qawolf)