- What’s new
- Script behavior changes
- List methods read every page
- Groovy: decimal literals are BigDecimal
- Groovy: getMonth() and getDayOfWeek() return enums
- Groovy: LocalDateTime.toString() drops zero seconds
- Python: an undefined name raises NameError
- Python: str + int raises TypeError
- Python: type(x) returns a type object
- Python: default argument values are evaluated once
- API whitelist
- What the administrator does
- If you are on 2.6.0
- After 5.0.0
JiBrok Studio 5.0.0 is a major version. It asks for two new Jira Service Management permissions, so Atlassian does not update existing installations on its own: a site administrator has to approve the update. Until then the site stays on the version it runs today (4.x or 2.6.0).
This page compares 5.0.0 with:
- 4.x - version 4.1.0 (a fix release on top of 4.0.0).
- 2.6.0 - builds of 2.6.0 made between 28.02.2026 and 24.03.2026; the exact build differs from site to site, so everything that changed after 28.02.2026 is treated as new for 2.6.0. If your site is still on 2.6.0, read If you are on 2.6.0 as well.
Some existing scripts return a different result after the update - without an error - or fail where they used to succeed. Check scripts that use these patterns, ideally on a test site first:
- List methods (
issue.getComments(),Users.find,Groups.findand others) now read every page, not only the first one. - Groovy: decimal literals are
BigDecimal;getMonth()andgetDayOfWeek()return enums;LocalDateTime.toString()drops zero seconds. - Python: an undefined name raises
NameError;str + intraisesTypeError;type(x)returns a type object; default argument values are evaluated once. - If your API whitelist was ever saved,
JSMcalls and attachment copies fail withblocked by whitelistuntil you add the rules from API whitelist.
Details and fixes: Script behavior changes. Steps for the update: What the administrator does.
What’s new
Script API
JSMnamespace - 23 methods for Jira Service Management: service desks, queues, request types, customer requests, public and internal comments, SLA, participants, approvals, organizations and customers. See Scripting API: JSM.Attachmentsnamespace -Attachments.list(),Attachments.copy()andissue.copyAttachmentsTo()with per-file, per-call and file-count limits. See Attachments. Read API whitelist before usingcopy: it needs a whitelist entry.- Field context options in
Fields- add and manage options of select-list fields. Issues.create()returns a chainable issue -await Issues.create(...).addComment(...).- Lists return every page -
issue.getComments()and the list methods ofUsers,Groups,Projects,Boards,Labels,Statusesand others now read all pages instead of the first one. This is a fix, but it can change results - see List methods read every page. - Message TTL for queues (
ttlMsinpush/pushFor).
Engines
- Groovy: operator overloading, switch expressions,
@Immutable,@Memoized,@Delegate, realBigDecimalandBigIntegerarithmetic. - Python:
import,functools.reduce,bytes,float("nan")/float("inf"), insertion-ordereddict,@classmethodcalled through an instance.
These bring both engines closer to real Groovy and Python. Some of it changes what existing scripts return - see Script behavior changes.
Triggers
- Custom JQL functions -
issue in jbExec("name", ...)runs a Studio script as a JQL function. Configure it on the new JQL Functions tab; results are precomputed on the five-minute tick. See Custom JQL Functions. - Scheduled trigger over JQL - run a script for each issue of a JQL query, optionally only for issues updated since the last run.
- Jira expression failures of workflow conditions now appear in the Error Monitor.
- Trigger parameters reach scenarios.
Scripted fields
- Four new field types: user, user list, group, group list.
- Values are written to Jira in bulk. If Jira rejects one value, the other fields of the issue are still written: the rejected field is retried on its own and stays empty.
See Scripted Fields.
Recipes and templates
- Bulk recipes by JQL, each with a preview: Bulk Update Field by JQL, Bulk Transition by JQL, Copy Field Values by JQL, Bulk Add/Remove Label, Bulk Set Resolution, Bulk Delete Issues by JQL (see Recipes).
- Two-project sync on the Recipes tab: manual, scheduled and event-driven runs; issue type, status and field mapping; comments, attachments, links and sub-tasks; JSM organizations, participants, approvals and SLA; dead-letter list, preflight check, export and import of sync configurations. See Two-project sync.
- New script templates in the template picker; a template opens with its own engine.
Console, Library and Admin Panel
- Rename a script in the editor title; see where a script is used before deleting it.
- Autocomplete for tables, queues, scripts, global variables and the script’s own variables.
- A warning when a script with a scheduled trigger is set to Run as: Current user (see Who a script runs as).
- Scenario output shows the time of every batch page; Cancel shows that it is working.
- The script form opens without flicker.
- Health Check runs all checks with one button; new Scenarios and Jobs panel.
- Export and import of a folder as a zip bundle, keeping script IDs.
- Export is no longer limited to 200 scripts.
- Issue links extracted from script and folder descriptions; links to deleted issues are cleaned up by Health Check.
- The interface uses Atlassian design tokens and buttons, so it follows the Jira theme more closely.
Not in this version
A REST API for keeping scripts in git and pushing them from a CI pipeline, and running exposed scripts from external AI clients over MCP, are in preparation and are not part of 5.0.0. Until then, scripts move between sites through Import & Export, and exposed scripts run through the Rovo agent.
Reliability
- Event triggers no longer queue Jira events that no script listens to (already in 4.1.0).
- Recipes and scenarios survive an app update in the middle of a run.
- More robust storage access and database migrations.
- Jira Expressions scripts are saved with their engine intact. Before, the engine name of a Jira Expressions script was cut short on save, so the script could not be picked for a workflow condition rule. The update repairs scripts and versions already saved this way; nothing to do on your side.
Script behavior changes
Nothing was removed or renamed in the Script API, and engine limits did not change. The changes below can still make an existing script return a different result without an error, or fail where it used to succeed. Please check scripts that match these patterns, ideally on a test site first.
List methods read every page
Before, a list method made one request and returned the first page (often 20-50 items). Now it walks pages until the end, up to 20 requests per call. Passing startAt or maxResults keeps the old single-request behavior.
Affected: issue.getComments(), Users.find, Users.findAssignable, Groups.find, Groups.getMembers, Groups.getBulk, Projects.list, Fields.getContexts, Labels.list, Boards.list, Boards.getSprints, Statuses.list, FieldConfigurations.getFields, FieldConfigSchemes.getMappings, FieldConfigSchemes.getProjects.
const groups = await Groups.find();
groups.length
// 4.x: 20 (first page only)
// 5.0: every group on the site
What to check: scripts that loop over these results and do something per item now do it for more items, use more of the per-run API call budget (maxApiCalls, see Limits) and take longer. On a large site pass maxResults if the first page is all you need.
Groovy: decimal literals are BigDecimal
def x = 1.5
x instanceof Double // 4.x: true 5.0: false
x instanceof BigDecimal // 4.x: false 5.0: true
(0.1 + 0.2).toString() // 4.x: "0.30000000000000004"
// 5.0: "0.3"
What to check: instanceof Double / instanceof Float on literals. instanceof Number works in both versions.
Groovy: getMonth() and getDayOfWeek() return enums
def d = LocalDate.of(2026, 3, 5)
d.getMonth() // 4.x: 3 5.0: MARCH
d.getMonth() == 3 // 4.x: true 5.0: false
d.getMonth() == Month.MARCH // 4.x: false 5.0: true
d.getDayOfWeek() // 4.x: 4 5.0: THURSDAY
What to check: comparisons of these values with numbers. d.getMonthValue() returns the number in both versions; d.getMonth().getValue() and d.getDayOfWeek().getValue() work in 5.0.
Groovy: LocalDateTime.toString() drops zero seconds
LocalDateTime.of(2026, 3, 5, 10, 0).toString()
// 4.x: "2026-03-05T10:00:00"
// 5.0: "2026-03-05T10:00" (same as Java)
What to check: scripts that compare or parse this string. Use an explicit format, for example DateTimeFormatter.ofPattern("yyyy-MM-dd'T'HH:mm:ss").
Python: an undefined name raises NameError
x = foo
# 4.x: x is None, the script goes on
# 5.0: NameError: name 'foo' is not defined
What to check: typos in variable names that used to pass silently, and code that tested whether a name exists by reading it.
Python: str + int raises TypeError
"Count: " + 5
# 4.x: "Count: 5"
# 5.0: TypeError: can only concatenate str (not "int") to str
What to check: string building with numbers. Use "Count: " + str(5) or an f-string.
Python: type(x) returns a type object
str(type("s")) # 4.x: "str" 5.0: "<class 'str'>"
type("s") == "str" # 4.x: True 5.0: False
type("s").__name__ # 4.x: None 5.0: "str"
What to check: comparisons of type(x) with a string. Use isinstance(x, str) (works in both versions) or type(x).__name__ == "str".
Python: default argument values are evaluated once
def add(item, acc=[]):
acc.append(item)
return acc
add(1)
add(2)
# 4.x: [2] (a fresh list on every call)
# 5.0: [1, 2] (one shared list, as in real Python)
What to check: functions with a mutable default ([], {}). Use acc=None and create the list inside the function.
API whitelist
Studio checks every Jira REST call a script makes against the admin API whitelist (Admin Panel → Admin → API Whitelist, see API Restrictions). The whitelist is on by default.
While a site has never saved it, Studio uses the built-in default list. Once an administrator changes anything on that screen (adds or deletes a rule, switches it off and on, presses Reset to Defaults), the whole list is saved and used from then on, and an update of the app does not change it.
When the saved list lacks rules of the current default list, the API Whitelist tab shows them in the warning “The default list has rules that your list does not” with an Add missing rules button, which adds only those rules and keeps your own. A call blocked for this reason also says in its error that the default list allows it.
JSM
The 5.0 default list adds the JSM namespace paths. A site with a saved whitelist keeps its old list, and every JSM call fails with blocked by whitelist until the administrator adds:
| Method | Pattern |
|---|---|
GET |
/rest/servicedeskapi/** |
POST |
/rest/servicedeskapi/request |
POST |
/rest/servicedeskapi/request/** |
PUT |
/rest/servicedeskapi/request/** |
DELETE |
/rest/servicedeskapi/request/*/participant |
Only the first line is needed for read-only use of JSM.
Attachments
The 5.0 default list adds the attachment download path used by Attachments.copy() and issue.copyAttachmentsTo(). A site with a saved whitelist keeps its old list, and both calls fail with blocked by whitelist until the administrator adds:
| Method | Pattern |
|---|---|
GET |
/rest/api/3/attachment/** |
The upload path (POST /rest/api/3/issue/**) and the metadata read (GET /rest/api/3/issue/**) are already in the default list. Attachments.list() needs nothing extra.
How to add a rule
Open Admin Panel → Admin → API Whitelist, choose the method, enter the pattern and press Add - the rule is saved at once. Add missing rules adds all the missing default lines in one step.
Reset to Defaults also brings in the JSM and Attachments lines, but it replaces every custom rule of the site with the default list.
If the admin whitelist is switched off on your site, scripts are not filtered and nothing needs to be done.
What the administrator does
Approve the two new permissions
5.0.0 adds two Jira Service Management scopes, used by the JSM namespace and by the JSM part of two-project sync:
| Scope | Why |
|---|---|
write:servicedesk-request |
Create customer requests, transition them, add participants, write request comments, answer approvals |
manage:servicedesk-customer |
Look up JSM customers and organizations |
Nothing else changed in permissions: no new access outside your Jira site, no change to data egress. Full list and reasons: Permissions Reference.
Update the app
Because the permissions changed, 5.0.0 is a major version and is not installed automatically. In Jira go to Settings → Apps → Manage apps, find JiBrok Studio, choose Update and accept the new permissions.
Database changes are additive and are applied automatically after the update; scripts, triggers and settings stay as they are.
Who a script runs as
The run-as model is the same as in 4.x; 5.0.0 makes it visible in more places:
- Run As settings in the Admin Panel decide which identities scripts may use: the current user, the app, a list of pre-configured users, or any user (see Configure execution identity).
- A scheduled trigger has no user session. A script set to Run as: Current user runs on schedule as the app and sees the issues the app can see. 5.0 shows this warning in the script’s Config tab; the behavior itself is unchanged.
- A script used as a JQL function needs a fixed identity (the app or a specific user) - “Current user” is rejected on save.
- Scripts running as the app cannot send
DELETErequests, send issue notifications, create groups or add users to them, or change project roles, workflow schemes, permission schemes or notification schemes.
Checklist
- Approve the update and the two JSM permissions (Approve the two new permissions, Update the app).
- Add the whitelist rules from API whitelist if your scripts use
JSMor copy attachments. - Look through scripts that match Script behavior changes.
If you are on 2.6.0
Sites on 2.6.0 did not approve the permissions of versions 3 and 4 and stayed on 2.6.0. Updating brings everything in the sections above plus the changes below.
More new permissions
Besides the two JSM scopes, 5.0.0 asks a 2.6.0 site for up to eight more (the exact set depends on the 2.6.0 build):
- seven JSM Assets scopes for the
Assetsnamespace:read:cmdb-object:jira,write:cmdb-object:jira,delete:cmdb-object:jira,read:cmdb-schema:jira,read:cmdb-type:jira,read:cmdb-attribute:jira,read:cmdb-icon:jira; read:screen:jirafor managing scripted field placement on screens.
What is new since 2.6.0
- The Admin Panel moved to a new interface, with the settings described in Admin Settings.
- Workflow validators and conditions written as scripts or Jira expressions.
- Incoming webhooks as a trigger.
- Scenarios with
loopandbatchsteps, and a sandbox mode for trying them. - Event triggers processed through a queue.
- Error Monitor, compliance baselines and drift detection.
- A stricter sandbox in all three engines, and higher script size limits.
Behavior changes since 2.6.0
These come in addition to Script behavior changes and apply if your 2.6.0 build is older than the date given.
eval() accepts only a script ID (13.03.2026).
await eval("1 + 1")
// 2.6.0 (older builds): 2
// 5.0: error - eval() accepts only the ID of a saved script
// works in both:
await eval("550e8400-e29b-41d4-a716-446655440000", { x: 10 })
Move inline code to a saved script and call it by ID.
Issues.search() returns all fields and pages automatically (04.03.2026).
const r = await Issues.search("project = ABC", { maxResults: 500 });
// 2.6.0 (older builds): without `fields` Jira returns only issue IDs;
// one page of at most 100
// 5.0: all fields when `fields` is not given;
// maxResults above 100 is fetched page by page
Pass fields to keep responses small.
Recipe renamed: backfill-scripted-fields is now Recalculate Fields (recalculate-fields, 14.03.2026).
Export file format: export now writes formatVersion: 1 (05.03.2026); an export made by an older 2.6.0 build with formatVersion: 3 is not accepted by the 5.0 import. Export again after the update if you need a backup.
Whitelist: a whitelist saved on 2.6.0 also lacks the field, screen and field configuration entries added to the default list in version 4. Add the ones your scripts need (Add missing rules adds every missing default line at once), or press Reset to Defaults if you have no custom rules:
/rest/api/3/field/**/rest/api/3/screens/**/rest/api/3/screenscheme/**/rest/api/3/fieldconfiguration/**/rest/api/3/fieldconfigurationscheme/**
After 5.0.0
These changes came after 5.0.0 and are part of the update of 15.11.2026.
Fixed
- Event scripts fire right after you save them. Before, a newly saved event script could miss events of its type for up to a minute.
- A scripted field whose Jira custom field was deleted is skipped. Its configuration is turned off; the other scripted fields and scripts of the issue are still processed.
- First launch with several administrators at once. Everyone who opens Studio while the database is being prepared sees Setting up database - this may take a couple of minutes on first launch... and waits; the error Failed to initialize database no longer appears.
- Thrown errors read as errors.
throw new Error("x")is shown asError: x, not as JSON. - Console notifications show their text. Saving a script without a name reads Failed to save script / Name is required instead of the internal message names.
Interface
- Every icon-only button shows its name on hover: the attention bell and Refresh in Compliance, Move up / Move down in the custom sanitize rules of a recipe.
- Global Variables: the editor and the Test Results / Version History column split the width evenly and stay inside the tab at every window width.
- Audit Log → Settings: the date stays on one line, Action and Category are as wide as their labels, and Details takes the remaining width. Script entries lead with the script name; the script ID is shown below it. Deleting a script records its name, so a Delete Script entry names the script.
- Console: leaving the Console tab with unsaved edits asks to discard them in a window titled Unsaved changes (it was titled New Script).
- Studio Log on the issue: the panel name is shown once - collapsed, the panel reads Studio Log Script runs (it read Studio Log twice), and the open panel no longer repeats Studio Log as a heading above the run list.
- Admin → JQL Functions: the tab description shows the JQL clause
issue in jbExec("yourFunctionName", ...)as code instead of text wrapped in backticks. - Admin → Script Usage: a usage of a deleted script shows the name the script had (when its runs are still in the audit log) next to the Deleted label, and the counter reads Total: 3 records (Total: 1 record for one).
- Admin Panel and Console - look: every tab and sub-tab shows its description at the same distance below the tab line, then its actions, then its content; section headings have one size and sit closer to their own description; the support, documentation, Console and Admin icons moved onto the tab line, so there is no empty band above the tabs. In the Console, the Output, Run History, Version History and Config panels start at the same distance below their tab line as the main tabs. Empty lists show one kind of empty state.
- Scenarios has a Refresh button instead of a bare arrow and, like Console and Library, opens with a short description of what a scenario is and where its runs are.
- Daily sync hour and the If entity already exists choice of Import are Atlassian selects; the section choices in Export / Import and the Trigger Types of error notifications are switches instead of browser checkboxes.
- The UIM counter reads 5 UIMs in Jira (it read 5 UIM(S) IN JIRA); buttons that start a row no longer look shifted to the right; the ✓ / ✗ mark of the Console Output stays on the status line.
Diagnostic Logging
While diagnostic logging is on, the app keeps the logs itself for JiBrok support. The interface no longer shows Forge CLI commands for reading logs.
