Data Processing Agreement

Last updated: October 31, 2026

Between

[Customer legal name], [address] ("Controller")

and

Helkrypt AI AS, org. no. 936 351 034, Postboks 14, 4756 Hovden ("Processor")


1. Subject matter and duration

The Processor provides the Atlassian Marketplace app(s) listed in Annex 1 to the Controller. This agreement applies for as long as the Controller has the app installed, and ends automatically when the app is uninstalled from all of the Controller's Atlassian sites.

2. Nature and purpose of processing

The app runs on Atlassian's Forge platform inside the Controller's Atlassian site. It processes data only to provide the app's features, as described in Annex 1.

3. Types of personal data and data subjects

See Annex 1.

4. Instructions

The Processor processes personal data only on the Controller's documented instructions. The Controller's configuration and use of the app are its instructions. If EU or EEA law requires other processing, the Processor will inform the Controller first, unless the law forbids it.

5. Confidentiality

Everyone at the Processor who can access personal data is bound by confidentiality.

6. Security

The Processor takes the measures required by GDPR art. 32. They are listed in Annex 2.

7. Sub-processors

The Controller gives general authorisation to use the sub-processors in Annex 3. The Processor will give at least 30 days' notice before adding or replacing a sub-processor, by email to the Controller's technical contact on the Atlassian Marketplace. The Controller can object within that period. If the parties can't agree, the Controller can uninstall the app, and the Processor will refund any prepaid fees for the remaining period. The Processor imposes the same data protection obligations on its sub-processors and remains responsible for them.

8. Assistance

The Processor helps the Controller, as far as possible given the nature of the processing, to:

  • answer requests from data subjects (GDPR art. 15–22)
  • meet its obligations on security, personal data breaches, data protection impact assessments and prior consultation (GDPR art. 32–36)

9. Personal data breaches

The Processor notifies the Controller without undue delay, and no later than 48 hours after becoming aware of a personal data breach that affects the Controller's data. The notice includes what the Processor knows at that time.

10. Deletion

When the app is uninstalled, the app data in Forge hosted storage is deleted by Atlassian according to the Forge hosted storage lifecycle. The Processor keeps no copies outside Atlassian. For Epic Rollup for Jira, the epic totals stored on epics (numbers only) are not removed by uninstalling. To remove them, an administrator turns off searchable totals and lets the background job finish before uninstalling.

11. Audits

The Processor makes available all information needed to show compliance with this agreement, and allows for and contributes to audits by the Controller or an auditor it chooses, with reasonable notice. The Controller covers its own audit costs.

12. International transfers

The apps send no personal data outside Atlassian. The optional AI features in Release Studio for Jira and Clear Transitions for Jira use Atlassian-hosted models (Forge LLMs), which run inside Atlassian. Any transfer by Atlassian is covered by Atlassian's own data processing terms with the Controller.

13. Governing law

Norwegian law. Disputes go to Oslo District Court (Oslo tingrett).

Annex 1 – Processing details

Auto Issue & Description Templates for JiraEpic Rollup for JiraSmart Defaults for JiraComment & Reply Templates for JiraRelease Studio for JiraClear Transitions for JiraSprint Trail for JiraClear Tests for Jira
PurposeAdd description templates to Jira work itemsShow progress totals (story points, time, percent done) for Jira epicsFill empty system fields (priority, labels, components, due date) on new Jira work items with default values set by an administratorLet Jira users post reusable comment and reply templates, written by an administrator, as their own commentsLet Jira users create release notes from versions with templates per audience, optionally improved or translated by Atlassian's built-in AI with source checking, publish them to Confluence, and keep frozen, approved release records as evidenceLet Jira administrators block workflow transitions until clear conditions are met, show users why, and optionally keep an anonymous block logLet Jira users search sprint history with JQL, see what changed in each sprint, and see the reasons found in the work item history for work that was carried over or removedLet Jira users write test cases, run them step by step in test cycles, create bugs from failed steps, link tests to work items and show test coverage, with optional AI drafts using Atlassian's built-in AI and test reports sent from the Controller's own CI
Data stored by the appTemplate names and text, selected project and work type IDs, default flag, last-updated timeAdmin settings (story points field per project, feature switches). Epic totals as a property on each epic: counts, story points, logged and remaining time, percent done, last-updated time. Numbers and a date only, no issue keys, titles or names.Rules: name, project and work type IDs, priority and component IDs, labels, number of days until the due dateTemplates: name, text, project and work type IDs, default visibility in service projectsTemplates, rules and settings; release records (release notes text, project, versions, template name, work item keys and status, time, checksum); short-lived AI resultsRules and settings; a rule usage list; an optional block log (work item key, rule ID, transition ID, time)A small property on each work item that has sprint history (sprint IDs, sprint dates as days, counts, last-updated time); sprint ID, name, state, dates and board ID; administrator settings; status of the history buildTests, folders, version history, links to work item IDs, cycles, results per step, actual results, comments and CI runs with the result log (Forge SQL); settings, one-way hashed CI tokens, a monthly AI counter and short-lived records, including a per-user pointer to unsaved AI drafts and, for 30 days, a marker with a Jira bug ID and key (no personal data) (Forge hosted storage); a small property on each work item with linked tests (number of tests, counts by status, time updated)
Personal dataNone stored by the app. Templates may contain personal data if an administrator writes it. The app reads reporter and assignee display names to fill in variables, and does not store them.None stored by the app. The app reads child issues (key, summary, status, story points, time fields, parent, team) to calculate totals, and does not store them. If an administrator turns on the optional "time logged per person" setting (off by default), the app reads worklog authors and logged time in the viewer's browser to show them on the epic. This is never stored.None stored by the app. The app reads the new work item's project, work type, priority, labels, components and due date to see which fields are empty, and does not store them. It reads no user fields.None stored by the app. The app reads the work item's key, summary, project, work type, reporter and assignee in the user's browser to fill in template variables, and does not store them. The posted comment is stored by Jira, not by the app.Atlassian account IDs of the users who froze or approved a release record. Free text in work item titles, templates and release notes may contain personal data that the Controller's users have written. Sent to Atlassian's built-in AI on request: work item key, text, work type, labels, components (no names)None by design: no account IDs are stored. Free text in rule names and messages may contain personal data that the Controller's administrators have written. The optional logged rule type evaluates the rule as the app, inside Atlassian, and stores no person data. Sent to Atlassian's built-in AI on request: the administrator's text and field, work type and link type names (no work item content)None by design: no account IDs and no names are stored, and the app does not read or store who made a change. Sprint names are free text that the Controller's teams write and may contain personal data. Work item key, summary and status are read to show reasons and are not stored. The app has no AI service and sends no data to Atlassian's built-in AIAtlassian account IDs of the people who created or changed an item, were assigned a test or ran it, and of the person who started an AI job. Names and avatars are not stored. Free text in tests, comments, results and CI messages may contain personal data that the Controller's users have written. Sent to Atlassian's built-in AI on request: work item summary, description and acceptance criteria, test text, failed test results and CI failure messages (no people fields, names or account IDs)
Data subjectsThe Controller's Jira usersThe Controller's Jira usersThe Controller's Jira users (only indirectly: work items they create get default field values)The Controller's Jira users and, in service projects, the Controller's customers (only indirectly: their names can be filled into a comment the user posts)The Controller's Jira users; people named in work item text (not by design)The Controller's Jira administrators (as authors of free text)The Controller's Jira users (only indirectly: people named in sprint names or work item summaries, not by design)The Controller's Jira users; people named in test text, comments or CI messages (not by design)
WhereAtlassian Forge hosted storage, in the same region as the Controller's Jira siteAdmin settings: Atlassian Forge hosted storage, in the same region as the Controller's Jira site. Epic totals: a property on the epic in the Controller's Jira site.Atlassian Forge hosted storage, in the same region as the Controller's Jira siteAtlassian Forge hosted storage, in the same region as the Controller's Jira siteAtlassian Forge hosted storage and Forge LLMs, inside Atlassian, in the same region as the Controller's Jira site where Atlassian offers itAtlassian Forge hosted storage and Forge LLMs, inside Atlassian, in the same region as the Controller's Jira site where Atlassian offers itAtlassian Forge hosted storage and the work item property in the Controller's Jira site, in the same region as the Controller's Jira siteAtlassian Forge hosted storage, Forge SQL and Forge LLMs, inside Atlassian, in the same region as the Controller's Jira site where Atlassian offers it; the work item property in the Controller's Jira site
RetentionUntil an administrator deletes the template, or until the app is uninstalledAdmin settings: until the app is uninstalled. Epic totals: until an administrator turns off searchable totals (a background job then removes them from every epic), or until the epic is deleted. Uninstalling does not remove them.Until an administrator deletes the rule, or the app is uninstalledUntil an administrator deletes the template, or the app is uninstalledRecords: the period set by the Controller's administrator (default 7 years). Templates: until deleted or uninstall. AI results: at most about 25 hours (deleted when fetched by the Releases page, otherwise by a daily cleanup of results older than one hour)Rules and settings: until deleted or uninstall. Block log: 90 days by default (30–365 by the Controller's choice)The work item property: as long as the work item exists and the app is installed. The app removes it from all work items when it is uninstalled, and an administrator can remove it earlier. Forge hosted storage: until the app is uninstalledTests, cycles and results: until the Controller deletes them or the app is uninstalled; deleted tests leave the trash after 30 days. CI run history and result log: as set by an administrator (default: keep everything); the latest CI result of each test stays on its cycle until the cycle is deleted. Short-lived records: 24 hours (30 days for the bug marker). The work item property: as long as the work item exists and the app is installed; the app removes it from all work items when it is uninstalled, and an administrator can remove it earlier

Annex 2 – Security measures

All apps

  • The app runs only on Atlassian's Forge platform ("Runs on Atlassian"). There are no Helkrypt servers.
  • No egress: the app sends no data outside Atlassian.
  • Helkrypt's developer accounts use two-step verification.
  • Atlassian's security measures for Forge apply to hosting and storage.

Auto Issue & Description Templates for Jira

  • Minimal permissions: read:jira-work, write:jira-work, storage:app.
  • Only Jira administrators can create, change or delete templates. This is checked on every request.
  • Logs contain only work item keys, template IDs and status codes. Never content or names.

Epic Rollup for Jira

  • Minimal permissions: read:jira-work, write:jira-work, storage:app, read:board-scope:jira-software, read:project:jira, read:epic:jira-software.
  • Only Jira administrators can change settings. This is checked on every request.
  • "Time logged per person" is off by default and is calculated in the viewer's browser. It is never stored, and is not shown on the dashboard gadget or in JQL.
  • Logs contain only work item IDs, event types, counts and status codes. Never content or names.

Smart Defaults for Jira

  • Minimal permissions: read:jira-work, write:jira-work, storage:app.
  • Only Jira administrators can create, change or delete rules. This is checked on every request.
  • The app reads no user fields.
  • Logs contain only work item keys, rule IDs, field names and status codes. Never field values or names.

Comment & Reply Templates for Jira

  • Minimal permissions: read:jira-work, write:jira-work, storage:app.
  • Only Jira administrators can create, change or delete templates. This is checked on every request.
  • Comments are posted as the user, so Jira's own permissions apply.
  • The app writes no logs.

Clear Transitions for Jira

  • Minimal permissions: read:jira-work, read:jira-user, storage:app. No admin scopes.
  • Only Jira administrators can create, change or delete rules and settings.
  • AI is optional and runs only when an administrator selects "Draft with AI". It uses Atlassian-hosted models (Forge LLMs) and is sent no work item content and no names of people.
  • The app stores no account IDs. The optional block log has no person data.
  • Logs contain only fixed technical keys and counts. Never work item content, rule messages, AI text or names.

Sprint Trail for Jira

  • Minimal permissions: read work items, write the app's own work item property, and app storage. No admin scopes.
  • Only Jira administrators can build or remove sprint history and change settings.
  • The app has no AI service of its own. The Rovo agent runs in the Controller's Atlassian site and the app's actions are read-only.
  • The app stores no account IDs and no names.
  • Logs contain only fixed technical keys, IDs and counts. Never work item content, sprint names or names of people.

Clear Tests for Jira

  • Minimal permissions: read:jira-work, write:jira-work, storage:app, report:personal-data. No admin scopes.
  • Views and the Rovo agent run as the user, so Jira's own permissions apply. Tests and results are checked against the user's Jira permissions in the project on every request.
  • CI reports are received at an address hosted by Atlassian. Each token belongs to one project, is stored only as a one-way hash, and is shown once. Report size and call rate are limited, and a report never creates Jira work items.
  • AI is optional. It runs only when a user selects an AI action, uses Atlassian-hosted models (Forge LLMs), is sent no people fields, and only makes drafts that a person accepts. An administrator can turn AI off or set a monthly limit.
  • The app stores Atlassian account IDs and reports them every week to Atlassian's personal data reporting service. It clears the account ID in the data when Atlassian reports that the account is closed.
  • Logs contain only the type of operation, IDs, counts, status codes and fixed error keys. Never test text, work item text, tokens or account IDs.

Release Studio for Jira

  • Minimal permissions: read:jira-work, storage:app, read:space:confluence, write:page:confluence, report:personal-data.
  • The weekly personal data report is made by the app itself and returns only the status of each account ID.
  • Confluence pages are created as the user, so Confluence's own permissions apply.
  • The text and export of a release record are shown only to the person who froze it, project administrators and Jira administrators.
  • AI is optional and can be turned off by an administrator. It uses Atlassian-hosted models (Forge LLMs) and is sent no names, reporters, assignees or links.
  • Logs contain only operation names, job IDs, status codes and AI token counts. Never content, names or account IDs.

Annex 3 – Sub-processors

Sub-processorPurposeLocation
Atlassian (Forge platform)Hosting, storage, logsSame region as the Controller's Atlassian site
Atlassian (Forge LLMs)Release Studio for Jira, Clear Transitions for Jira and Clear Tests for Jira only: optional AI features, using Atlassian-hosted modelsInside Atlassian, in the same region as the Controller's Jira site where Atlassian offers it

Signed for the Controller: ______________________ Date: __________

Signed for Helkrypt AI AS: ______________________ Date: __________