Managing fieldwork
Once you have tested a Script version, create a Run to collect responses from it. Choose how people enter, review the collection settings, then activate it. QubeScript and Odin use the same Run workflow.
Open Participants or Runs in the QubeSurvey app. Respondent links open separately from the workspace.
Each accepted, retained successful response costs US$0.10 for USD Accounts and is charged to prepaid credit. Use invented participants and answers while testing.
On this page
- Endings and retained responses
- The objects you work with
- Prepare Participants and a Group
- Prepare a Run
- Draft email messages
- End-of-survey destinations
- Start, resume and change a version
- Pause, resume or close
- Configure quotas
- Inspect and change quota targets
- Configure quota priority scores
- Inspect responses
- Create a private answer export
- Work from the command line
Endings and retained responses
A new Run needs no outcome setup. The ending named complete keeps the answers and is
charged as an accepted response. Every other ending keeps nothing: the case is still
recorded (for example as screened out, quota full or refused), no data is returned and
nothing is charged. The names screenout, quota_full and refused are recorded
under those meanings; any other name is recorded as other.
To keep the answers of another ending, open Settings → Interview outcomes and tick Keep answers (charged) for it, then choose Save outcomes. Kept answers are billed at the same rate as a completion. Changes apply to new interviews only.
For Odin, a Run lists the response codes its script can end with. It takes them from
the literal codes in the script only: *ENDST 41 lists code 41, and *ENDST "041"
lists the different code 041. Codes 18, 19 and 22 are always listed. A survey.ini
file is not read on QubeSurvey, so its labels and response-code settings do not apply.
| Code | Default meaning | Answers |
|---|---|---|
18 (*END) | Complete. Fixed. | Kept and charged. |
19 (*ENDNGB) | Partial. | Kept and charged; you can untick it. |
22 (*ABORT) | Broken off. | Not kept; free. |
| 08 | Refused (NIPO’s default refusal code). | Not kept; free. |
| 16 | Screened out (NIPO’s “outside target group”). | Not kept; free. |
| 21 | Quota full (NIPO’s “stratification maximum reached”). | Not kept; free. |
| 23 | Broken off (NIPO’s “interview broken off”). | Not kept; free. |
any other *ENDST code | Other. | Not kept; free. |
Under Settings → Interview outcomes each code has a meaning to choose and a Keep answers (charged) box. Code 18 cannot be changed. A complete always keeps its answers. Save outcomes to apply them to interviews that start afterwards; interviews already started keep the settings they started with. The readiness list names each script code that still has the default “other, not kept” meaning, so you can map it before going live.
Every code that keeps answers is charged as an accepted response and is included in the Odin answer export. A code that does not keep answers is free and returns no data.
A script whose *ENDST argument is a variable or expression cannot run on a hosted
Run, because the codes cannot be listed in advance. The readiness list reports it and
activation refuses it. Use a literal code. A code must be 1 to 4 digits, and a script
can name at most 64 different codes.
Before an Odin interview starts, QubeSurvey checks that the script can collect responses. It checks submitted answers by running them through the Odin runtime. See the Odin reference for supported commands and host differences.
Progress is saved in the respondent’s browser. An already cached untimed interview can reopen offline on the same device while its 24-hour device cache remains current; each accepted save extends that retention window. Starting an interview and submitting its final response require a connection. An answer saved on the device is included in the Run only after QubeSurvey accepts the submission. If confirmation is missing, retry the saved submission. QubeSurvey can confirm an already accepted submission even after the Run closes. A new submission may be refused after closure, expiry or revoked access. Leaving a timed activity does not pause its countdown.
The objects you work with
| Object | Meaning |
|---|---|
| Account | Owns hosted Scripts, participants, Groups and services; carries the shared billing balance. Active membership and Script roles govern access. |
| Script | A questionnaire, its uploaded versions, tickets and collaborators. Older screens or APIs may call it a project. |
| Version | A saved copy of an uploaded script. It stays unchanged; a Run uses the selected version for new interviews. |
| Participant | A reusable account-owned person record with optional identifier, name and sample attributes. |
| Group | A reusable set of Participants, used for invitation or signed-in assignment cohorts. |
| Run | A round of data collection using a selected Script version, with its own access settings, responses and optional Services. |
| Attempt | One interview assigned to or started by a respondent. APIs may call it a case. A Participant can have more than one attempt. |
| Service | An AI task configured under Services. In Run Settings, select the published revision to use for each service name in the script. |
Write the questions, routing and answer checks in your script. In Runs, choose who can take the survey, which script version to use, and any AI services it needs. This is also where you monitor responses and download data.
Prepare Participants and a Group
- Open Participants from the product bar and choose the owning Account. Use People, access and billing if the Account or controls are missing.
- Choose Create group, enter its name and optional description, then open it.
- Choose Import sample and select a CSV, TSV or XLSX file of invented people. Choose the identifier column. If a sample schema is configured, review its stable field keys, column mapping, required values and types.
- Preview the import. Review accepted, duplicate and rejected rows before importing; a preview is not a saved participant cohort.
- Inspect the members and the import audit. Use Participant data to review or correct a member’s identifier, name and supported sample attributes.
To add one person without a file, open the Group and choose Add participant. With an active sample schema, enter its fields; otherwise choose an email address or external identifier and optionally enter a display name. The server applies the Group’s validation and duplicate checks, then saves one accepted participant. An email identifier is saved as an email contact point. No invitation is sent when a participant is added.
A Group is reusable across Runs. Removing a member is different from erasing their data. Participant erasure removes identity/contact data under the product’s erasure path; it is not a promise that every answer or previously downloaded file is anonymous. Questions may themselves collect identifying information.
An open survey does not require importing a Group first.
Prepare a Run
- In Versions, choose Collect responses on the Script version you already
tested. This creates a draft open-survey Run named after the Script and today’s
date, with the progress bar only and the human check off, and opens it. If a
draft Run for that version already exists, that Run opens instead. Creating a
draft does not open response access.
To set the name or other options first, open Runs, keep the intended
Account selected, choose the create-Run action and select the Script and
Version you already tested. Then:
- Choose Activity Type: CAWI survey, Practice quiz, or Timed activity (participant-visible services). For a quiz, configure the score variables and any feedback you want to show.
- Choose Access using the table below. Personal invitations and signed-in assignments require a Target Group; an open survey does not use one.
- For a timed activity, review the optional time limit. Choose the respondent progress display; it estimates questionnaire progress, not score or time left.
- Choose Create draft run.
- At the top of the draft Run, Ready to activate? lists, in order: the pinned version, the script’s endings, which endings keep answers, the credit available (as a number of accepted responses), the quota configuration, the opening dates and the Run’s status. Each item is ready, blocked or worth checking, with one sentence saying why and a link to the section that fixes it. It only reads; it changes nothing. The same list appears in the activation dialog.
- Open Settings to fix what the list shows. For Odin, Interview outcomes lists
the script’s response codes (see Endings and retained responses).
For QubeScript, Interview outcomes
are already set:
completekeeps its data and is charged; any other ending is free and returns no data unless you tick Keep answers (charged). Activation still requires the script to declare at least one@outcome. Zero credit is reported as a warning: the Run can go live, but its link refuses responses until credit is added. Review services if the script calls them; see Services. Quotas can be configured only while the Run is a draft. Once it is active, the quota panel reports that setup requires a new draft Run, so decide the targets before you activate — an active Run is not the place to add a cap. - Choose Activate run, review the current cohort/access and confirm. The button is disabled while any item is blocked. Activation allocates attempts for the current Group, or opens the shared link. Test the respondent path yourself before sharing its link.
| Access | How to use it |
|---|---|
| Personal invitations | A selected Group receives individually allocated attempts. Keep personal links private to their intended recipients. |
| Open survey | In Settings, copy Shared survey link. Each new visit can create a separate response; the link does not establish one response per person. |
| Signed-in assignments | Participants use Portal and must sign in with the verified account identity linked to their assignment. A forwarded link does not grant another user access. |
Survey email is sent only when you launch a message from the Messages view, after reviewing its recipients. If a message cannot be sent, that view states the reason. Activating a Run does not send email.
Draft email messages
Open a Run’s Messages view to create an invitation or reminder draft. The
message body supports paragraphs, bold, italic, list markers such as
- item, and links such as [Study information](https://example.org). Only
HTTP(S) links are active; raw HTML and images are not supported. Use
{{campaign_name}}, {{survey_link}} and {{opt_out_link}}; the Run name is
inserted as text and the survey and opt-out placeholders become links. The live
preview updates while you type, using the selected Run name and synthetic
survey/opt-out links rather than a recipient’s personal link. Saving a draft does
not send email. Preview and recipients shows the saved message and recipient
check only. Sending is available only once survey email is enabled; it costs
US$0.01 for each email accepted by the provider, as listed in
People, access and billing.
Check email status
Under a saved message, Delivery status shows preparation, sending attempts and provider reports. Refresh status only checks the recorded state; it does not send or resend email. Provider acceptance means the sending provider accepted the request. A delivery report means the recipient’s mail server accepted it, not that it reached the inbox or was read.
Temporary delays, bounces, failures, rejections and complaints appear when the provider reports them. Terminal outcomes take precedence over temporary delays. If sending is unconfirmed or delivery is unknown, check the original attempt before sending a replacement to avoid duplicates.
End-of-survey destinations
In a Run’s Settings, use End-of-survey destinations to send respondents to another page when their interview finishes. Enter a Default destination, or expand Outcome-specific destinations to use different URLs for completion, screening out, quota full, or another supported outcome. An outcome-specific URL wins over the default; with neither set, QubeSurvey shows its usual end screen.
Destinations work with QubeScript and Odin. They are selected from the accepted
server outcome, after the submission has been accepted. Failed submissions remain on
the retry screen. Use complete https:// or http:// URLs without embedded
usernames or passwords. QubeSurvey does not append answers or interview tokens.
Changes apply to interviews started after saving. Started interviews keep their original destinations. Named outcomes mapped to the same outcome type share its destination.
Start, resume and change a version
Open the survey link or Portal assignment. Opening the link displays the study’s welcome screen; the interview officially begins when the respondent chooses Start. For timed activities, the timer starts from that moment. Reopening a link resumes the existing attempt when permitted by the study settings.
Uploading a Script version does not change a Run or its started attempts. Each Run keeps the selected Script version. Create a new Run to collect responses with another version, then review its outcomes and test it before activation.
A Run with selected quota configuration also keeps its opening dates and quota definition. Create a new Run for a different quota definition.
See the commuter-travel study for a short example of screening, multiple answers and conditional follow-up questions.
Pause, resume or close
In Run Settings, use the Run status controls and review the confirmation.
- Pause run stops new starts, online reopening and fresh submissions until resumed. Already open interviews may continue offline on the device; pausing does not delete or expire those attempts.
- Resume run permits online access and submissions within the opening dates. Started attempts retain their original source and saved history.
- Close run cannot be undone. It stops new starts and fresh submissions permanently. Existing accepted history and retained answers remain. An open offline interview may have answers that can no longer be submitted.
Review started-attempt counts before closing. If another operator changes the Run while your confirmation is open, reload the current state before trying again.
Delete a run
An account owner or admin can permanently delete a Run in Settings → Delete run by typing the Run’s name. A Run that retains responses, exports or accounting (prepaid, participant service, quota or messages) asks for a second explicit confirmation and disposes of those records with the Run. This cannot be undone; to keep the records, Close the Run instead.
Configure quotas
Quotas limit how many respondents can continue, overall or in groups defined by their answers. For example, a study could aim for 250 responses overall and 100 from people in the North. A Hard quota can stop further entry when full; Soft and Observe cells track responses without blocking entry. Your script asks for a place after the relevant screening questions and decides what to do if none is available.
Quota setup is available in Run Settings for an unused open-link QubeScript draft before activation. It requires Account membership and Script edit access; the draft must have no Group or attempts, including allocated attempts. Existing active Runs and Odin Runs do not use this setup path.
- Open Settings on the draft, then Configure quotas. Review the selected immutable Script version and opening dates.
- Enter Quota binding name, matching the name in your script’s quota allocation call. Enter the overall Desired minimum, Nominal maximum, Extra admissions / tolerance and Reservation lease in seconds. The overall quota is hard and has a finite maximum. The lease sets how long a respondent’s place is reserved while they answer the rest of the survey.
- Choose Add quota cell for each answer-based cell. Give it a stable ID, label, capacity and Hard, Soft or Observe mode. A soft cell may have no maximum; Observe records counts with no target. Choose questions and exact answer values, or an inclusive range for an integer question. Multiple conditions in a cell use AND; values within One of use OR.
- Choose Review quota configuration. Check the binding, targets, modes, answer conditions and selected Script, then choose Configure quotas. Opening Settings or reviewing a form does not select a configuration.
- Check Interview outcomes if you want to keep answers from an ending other than
complete, then separately choose Activate run. Test a response that gets a place and one that reaches a full quota. Check the ending shown to each and the retained response data.
For example, a script using
environment.allocate("campaign-quota", "screening") needs the quota binding
name screening. It returns true when a place is reserved and false when
one is unavailable. Use that result to continue the survey or show the full-quota
ending. Define the cell conditions from questions already answered before
the call, and name the script’s full-quota ending quota_full so it is recorded as quota full.
Cell conditions, modes, binding, lease and selected Script become fixed once configuration is selected. Minimums, maxima and tolerance can change later. An overall-only quota needs no cells; cells can overlap and their capacities need not sum to the total. Choice conditions use the script’s typed values, not interchangeable labels.
If setup is awaiting confirmation, your browser keeps the exact reviewed command. Choose Retry quota setup to confirm its original result. Another operator can use Review quota recovery to inspect the original configuration and author before recovering it. Recovery can confirm earlier setup after activation or closure; it does not reset targets or reopen the Run. Pending activation has Retry Run activation. Resolve unsaved changes and pending commands before changing Run status.
Inspect and change quota targets
After confirmed setup, Quotas shows the current capacities and counts. An unavailable or uninitialized message is not an empty quota table.
The table shows targets, accepted completions and places reserved for ongoing interviews, both overall and for each cell. It also shows how many more places are available and how far completed responses fall short of the desired minimum. Open Conditions to see which answers put a respondent in a cell. Cells can overlap, so their counts and targets need not add up to the overall total. Counts are a loaded snapshot; Refresh quotas reads current state.
The desired minimum is a target for accepted completions. Reserved places do not count toward it. A hard quota allows entry up to its nominal maximum plus tolerance. For a maximum of 250 and tolerance of 10, 259 completed or reserved places leave one place available; 260 leave none.
A reservation holds a place for the configured lease period. Soft cells never block entry; Observe records counts without a target or entry limit. Completion counts use the final accepted answers, including changes made after Back or resume. A respondent can therefore finish in a different cell from the one that reserved their place, and completion counts can exceed a target. Lowering a target keeps existing reservations and accepted answers.
To change capacity, choose Edit targets, enter non-negative whole numbers and a Reason for changing targets, then Save capacity changes. For a finite maximum, the minimum must not exceed it. Unbounded cells have zero tolerance; Observe keeps minimum zero and an unbounded maximum. The full capacity set is saved against the loaded revision. Conditions, modes and lease remain fixed. A closed Run permits reading counts and recovering an earlier saved command, but refuses a new target change.
If another operator changes targets while you are editing, refresh to inspect the latest numbers, then select Review current revision before saving. Your proposed values and reason remain. If a save is awaiting confirmation, choose Retry target change to confirm the original command; do not replace it because the connection failed. The receipt records who changed targets, why, when and the before-and-after values; the table shows current state, which may include later changes.
Target management requires active Account membership and Script edit access. Aggregate dashboard access does not expose configuration or target editing. The platform supplies counts and recorded observations; the script decides how to use them. Expression and completed-cell rim score setup are described below. Completed-cell scores use accepted final answers; sample-inventory weighting is a separate capability.
Configure quota priority scores
Priority scores let your script compare which groups still need responses. For example, a score can be higher for a region with many places left than for one near its target. A score is advice for your survey logic; reserving a place still uses the separate quota allocation call.
After confirmed quota setup, Quota priority scores can bind a numeric expression or a completed-cell rim on an unused open-link QSS draft where quota collection is enabled. Up to 100 immutable bindings share the existing score capacity. Each binding has its own original actor, reason and receipt.
- In New score binding, enter Score binding name, matching the name your script will request.
- Select Expression in Score strategy, then enter Quota score expression
and a reason. Use
target(),reserved(),contributed()andremaining()for the overall quota, or supply an exact cell ID such asremaining("north"). Arithmetic,min()andmax()are available. This is a bounded expression language; it does not execute SQL. - Inspect Current quota reference for the actual cell IDs, targets, reservations and contributions. Choose Preview score to evaluate the exact expression against those current counts. The preview shows its raw value, score 0–1000 and target revision; it reserves nothing.
- Choose Review score binding, inspect the immutable definition, selected source/policy, preview and attribution, then Confirm score binding. Editing the expression or refreshing the reference requires a new preview.
Fresh creation independently checks the definition and target revision again before recording a binding. Invalid arithmetic, a value outside 0–1 or changed targets refuses creation. The preview uses current counts; later scores may change or refuse. Original admitted commands and binding history still recover without a new preview, including older commands with no target revision. Confirm every required binding before separately activating the Run.
For example, min(1, max(0, remaining("north") / max(1, target("north"))))
returns the remaining fraction of the North nominal maximum. target() and
remaining() require a finite maximum; references to unbounded cells are
refused, including in an unused expression branch. reserved() and
contributed() can read their counts. Remaining subtracts completed
contributions and active reservations, with a lower bound of zero. A fresh metric
must return 0..1; invalid, non-finite, out-of-range or division-by-zero results
record no score. Guard zero targets explicitly when dividing.
The script uses the existing service primitive:
function north_priority() returns integer = service.call("quota-score", "north-priority").scoreThe expression result contains integer score (0..1000) and target_revision,
and text policy_revision. Higher values mean higher configured priority. The script
owns thresholds, routing and study decisions. A score creates no reservation;
use the separate quota allocation call for admission. Even score 1000 cannot
override a full hard quota.
For Completed-cell rim, choose that Score strategy and enter Completed-cell rim configuration (JSON) using existing quota cell IDs. Dimensions are disjoint lists of categories; tuples declare explicit joint support in dimension order. For example:
{
"dimensions": [
{ "id": "age", "cellIds": ["young", "older"] },
{ "id": "region", "cellIds": ["north", "south"] }
],
"tuples": [
{ "id": "yn", "categories": ["young", "north"], "supported": true },
{ "id": "ys", "categories": ["young", "south"], "supported": true },
{ "id": "on", "categories": ["older", "north"], "supported": true },
{ "id": "os", "categories": ["older", "south"], "supported": true }
]
}Declare every allowed tuple. supported:false marks a structural zero;
an omitted tuple is not silently added. A supported empty tuple uses a positive
one-interview seed. Current final target margins are fitted to the canonical
joint completed tensor. The raw priority is fitted desired final completions /
max(actual completed, 1), then normalized to 0..1000. Counts may exceed targets;
the score neither reverses accepted work nor grants new capacity. When every
achieved margin is at least its target, actionable priority is zero.
Preview score shows readable dimensions, final targets, explicit support, fitted desired counts, raw priorities and normalized scores. Inspect them before Review score binding and Confirm score binding. Editing the definition, changing strategy or refreshing targets invalidates that preview and review. The dimensions and support become immutable with the binding; target revisions can change later. Unclassifiable completed data, unsupported requested tuples, revision exhaustion and a failed actionable fit refuse a fresh score. These advisory refusals do not authorize rejecting an otherwise valid completion.
Limits are 2–8 dimensions, 500 categories in total and 2000 declared tuples. Each completed command and receipt is bounded to 256 KiB; the Run’s combined score definitions are bounded to 1 MiB and share the 100-binding limit. Expression bindings keep their existing smaller bounds. A draft too large for safe device storage is refused before dispatch; it is not truncated.
The script requests one actual category from each dimension through the same primitive. The host independently checks the verified answers and tuple support:
function joint_priority() returns integer = service.call("quota-score", ["joint-priority", "young", "north"]).scoreThe literal example applies to the Young/North tuple. Derive the category IDs
from the interview’s actual answers. The returned struct contains integer
score, target_revision, plan_revision and contribution_revision, plus text
policy_revision and semantic_version (completed-target-rim-v1). These are
the recorded fit’s basis, not a promise that current counts still match it.
Scripts own study thresholds, routing and messaging; allocation remains separate.
Bindings become fixed at creation and cannot be added after activation. If activation wins a concurrent creation, that fresh creation is refused without late application. Admitted · unconfirmed means an original command was selected; it does not prove that a binding was accepted. Review recovery can confirm an original accepted binding after activation or closure and retains its original actor. It cannot create a new binding in that state.
If a score command is awaiting confirmation, the browser keeps its exact request. Choose Retry score command to recover its original result; do not replace a pending command after an unknown error. Resolve unsaved changes and pending commands before changing Run status or leaving the Run.
Scores are recorded per native call identity and repeat scope. Back, reload and replay reuse the original value, even after targets or completed counts change. The quota score service supports registered expression and completed-cell rim bindings. A new device can answer again using authorized observations; it does not recover unsaved answers from another device.
Inspect responses
Open Responses for aggregate counts and the authorized attempt roster. The UI shows the latest synced state; offline work may be absent until the respondent reconnects. Completed, finished and expired attempts are different statuses. History can include reset or erased attempts, so counts are not a unique-person census.
| Status | Meaning |
|---|---|
| Completed | Accepted successful completion with retained answers. |
| Finished | A terminal ending other than successful completion, such as screening out or refusal. Its saved outcome policy decides retention. |
| Expired | An attempt that can no longer be completed under its current access/lifecycle policy. |
Dashboard-only access exposes aggregate progress and withholds respondent details, operational controls and exports. Script operators have separate permissions. Changing a progress-role assignment does not grant Script editing.
What an export contains
An export is a snapshot taken when you create it. Cases that finish afterwards are not in that file; create a new export to include them. That also means two exports of the same Run can differ by a few minutes and a few rows.
- Values are codes, with the question wording and every code’s meaning in
dictionary.json, which is in the ZIP. That file is the metadata: analysis software reads it, a spreadsheet does not. If you need a file you can read directly, that is a separate option on the export screen rather than something the dictionary lacks. - Repeat blocks are keyed by option, not by order. Columns for a repeat over
selected options are
name__r1,name__r2, … where the suffix follows the option’s declared code, so__r2can be a different statement for a different respondent. Never read a repeat column as “the second thing this person chose”. - Every row carries a case identity, not a respondent identity. The first three
columns are
case_id,submitted_atandoutcome_name.outcome_nameis the ending the interview reached: the QubeScript ending’s name, or the Odin response code.case_idis a pseudonymous reconciliation key: it matches a row to one case in the Responses tab and to a support request, but it is not a name, email or device. Turning on participant columns on an attributable Run adds those fields after the identity columns; it never changes or removes them, and an anonymous export never gains a participant name or email. The file still has no per-case charge, so reconcile charges in the Responses tab or from your invoice.
Create a private answer export
For a Run with completed retained answers, open Exports:
- Choose Export version to set the questions, labels and order, or choose Union of versions. Choose All data or Only data collected using this version where available. A union includes all versions and leaves blanks for missing answers.
- Choose CSV, JSONL, or CSV (SSS). The current layout is wide. CSV (SSS) includes its Triple-S metadata; all formats include a data dictionary.
- For SSS, choose exact text or the explicit single-line conversion. Stored answers stay unchanged. CSV refuses formula-like text; use JSONL to preserve it.
- Choose Create export when enabled. Wait for Ready, then download from your own authorized session. Use Refresh if processing status is delayed.
The standard answer dataset includes completed retained responses. Retaining a finished unsuccessful attempt does not include it automatically in that dataset.
Exports are private to the requester and exclude sample identifiers by default. That does not remove identifying text from answers. An erasure or permission change can invalidate a prepared export. Treat any file you already downloaded according to your own retention obligations.
The current export limits are 1,000 responses, 8 MiB of response history, 100,000 answer cells and 16 MiB of output. An unsupported structure, lossy profile or exceeded limit refuses the export rather than creating a partial file. Queued exports depend on deployment configuration. Parquet and larger-volume streaming remain unavailable.
Odin responses export as wide CSV or JSONL with a data dictionary and exact text. Select one Odin version; the browser selects Only data collected using this version. Whole-run requests containing completed responses from another version refuse. Union, SSS and U/O files are unavailable for Odin. Only verified retained code18 completions belong to this dataset. Retained partial19 responses are excluded; older unpinned responses refuse rather than being silently omitted. Source eligibility and fixed analysis structure are checked again before export. The 8MiB limit includes the sealed server launch, policy and input evidence. Older responses without a server verification pin cannot be exported through this dataset.
Work from the command line
After qube login, the qube command can create and operate a Run and download its
data: qube run list, qube run show <run>, qube run create --version <id>,
qube run activate <run> --yes, qube run pause|resume|close <run>, and
qube export create <run> --format csv --wait. The same Account and Script
permissions apply as in the browser. activate and close do nothing without
--yes. See Use qube with an AI agent.