Privacy Policy
This policy explains what data the AccessRadar app for Jira Cloud processes, where it lives, why, and what you can do about it.
Last updated: 2026-11-12 · Version 1.6
On this page
Summary
- AccessRadar is a Forge app that Runs on Atlassian. All app data is stored in Atlassian's Forge hosted storage (Forge SQL), and it follows your Jira site's data residency.
- AccessRadar makes no calls to servers outside Atlassian, and Radrly receives none of your Jira data.
- Atlassian is the only sub-processor (hosting). AccessRadar calls no external AI or LLM service. If your admin turns on Atlassian's external MCP access for AccessRadar, a Jira admin can ask the AI client they connect with Atlassian sign-in for AccessRadar answers (see AI and Rovo).
- The app is read-only toward Jira: it never modifies work items, users, groups or permission schemes.
- Radrly acts as a data processor for the personal data in the app, and your organisation is the controller.
Who we are
AccessRadar is developed and operated by Radrly Sp. z o.o., a company established in Poland (“Radrly”, “we”). Registered office: ul. Tarczyńska 68, 05-831 Krakowiany, Poland. KRS 0001215510, NIP 5342706013, REGON 543658680.
This policy covers the AccessRadar app. “Customer” means the organisation that installs AccessRadar on its Jira Cloud site. “Users” means the people in the Customer's Jira site.
Data we process
AccessRadar processes only the data needed to resolve effective access, take snapshots, run reviews and produce exports.
| Data | Examples | Purpose |
|---|---|---|
| Atlassian account IDs | IDs of users and app accounts that hold roles, group membership or grants, and of people who decide, sign, grant an exception, create a campaign, are assigned a review or appear in the audit log | Attribute access, decisions and sign-off to the right person |
| User display names | The display name kept with each person record (together with account type and active status), and names quoted in in-app alert notices | Make reviews and alerts readable for admins and auditors |
| Group and role assignments | Group IDs and names, group memberships, space role IDs and actors, application roles | Resolve the path of access |
| Permission scheme data | Scheme IDs, holders and permissions; space-to-scheme mapping | Explain why a person has access |
| Snapshot and review records | Resolved access rows, change diffs, review items and decisions (Keep, Revoke, Exception), free-text notes, signer account ID, sign-off time and time zone, attestation | Access reviews and audit evidence |
| Signature chain hashes | SHA-256 evidence hash, coverage hash and previous-review hash of each signed review, its chain number, and snapshot content hashes | Tamper-evident sign-off and chain verification |
| Access exceptions | Subject, space or group, permissions, justification text, expiry date, status and the account ID of who granted it | Track accepted risk with an expiry |
| Campaigns and assignments | Campaign name, scope, schedule, assignee rule and reminder days, the creator's account ID; per-space assignments with the assignee's account ID, due date and status (Advanced) | Run recurring reviews and delegate them to space owners |
| Notices and remediation results | In-app change alerts and campaign reminders (type, severity, title, text, target space or account, who dismissed it) and the result of checking Revoke decisions on later snapshots (Advanced) | Show alerts and reminders in the app |
| Audit log | Account ID of the actor, action, target and counts, including each Rovo or MCP tool call (caller, action, row count) | Accountability for actions in the app |
| Export artefacts | CSV, PDF and JSON content generated in your browser for download (not stored by the app) | SOC 2 / ISO 27001 evidence packages |
Email addresses are not stored. The app does not request them. It does not store work item content, worklogs, comments, descriptions or attachments. Free-text notes, justifications and campaign names are typed by the Customer's users, so please do not put personal data in them.
From version 3.0, the Sharing and security page also reads, live and only for the signed-in Jira administrator: filter and dashboard names, owners and share targets; issue security scheme and level names, who is granted each level, and which spaces use the scheme; and, on Advanced, Jira audit-log summaries of permission, group and role changes (actor, object, changed values) for the last 7 days. AccessRadar does not store this data, does not display or keep filter JQL, and does not read work item content. Sites that have not approved the 3.0 update do not grant these scopes; that section shows an approval notice and the rest of the app works as before.
Who sees what in the app. Jira administrators can see all data above. Space owners who are assigned a review see only their own space's data and decisions, and cannot see other spaces.
We do not use this data for advertising, profiling or sale, and we do not combine it with other data.
Where data is stored
All AccessRadar data is stored in Atlassian Forge hosted storage (Forge SQL), on Atlassian's infrastructure. AccessRadar stores no End-User Data outside Atlassian products and services, and does not process it outside Atlassian or the user's browser.
Data residency. Forge SQL supports data residency. The app's data is kept in the location chosen for your Jira site and moves with it if you migrate between residency locations. Radrly does not run any servers that hold your data.
No egress, no sub-processors
AccessRadar declares no remote hosts, uses no Forge Remote and no Connect, and does not call external APIs. It exposes no public REST API of its own. The Rovo actions, also offered as Rovo and MCP tools, are Forge modules that run inside Atlassian; external access to them goes only through Atlassian's MCP service, with Atlassian sign-in and your admin's approval.
We do not share End-User Data with third parties. Atlassian provides the hosting platform (Forge, Jira and Rovo) under its own terms and, for GDPR purposes, is a sub-processor. Other than Atlassian, we use no sub-processors.
AI and Rovo
AccessRadar does not call any external AI or LLM service. Access resolution and risk indicators are deterministic calculations inside Forge.
On Advanced, a Jira administrator can press Suggest review order (AI) on an open review. AccessRadar then asks Atlassian Forge LLMs (the llm module, on Atlassian's platform) for an order suggestion. There is no third-party AI provider, no API key, no remote and no egress; the data stays on Atlassian. The call is optional and never runs on its own. Per item, for at most 120 items, the app sends a pseudonymous label, the access type (including group and role names), the existing risk class, the deactivated-account flag, and whether the item is new. It never sends display names, account ids, emails, space or scheme names, notes or issue data. Suggestions are held in the browser for that page view only. AccessRadar does not store them or use them for training. The audit log records only counts and the admin who pressed the button. This is advice only: the model never decides, and the reviewer decides every item.
The Advanced edition includes five read-only Rovo actions (who can access, why, person access, changes since, review status). They are used by the AccessRadar Rovo agent and are also offered as tools through Atlassian's Rovo MCP service. In Rovo they run under Atlassian's terms and AI trust policies.
External AI clients (off by default). A site or organisation admin can turn on external MCP access for AccessRadar in Atlassian Administration (Apps → Sites → your site → Connected apps). A user can then connect an MCP-compatible AI client of their choice (for example Claude or Cursor) to Atlassian's MCP endpoint and sign in with their Atlassian account. When that user asks, the tool answer (for example space keys, group names, user display names, permission keys and review status, up to 50 rows) is sent through Atlassian to that AI client. AccessRadar answers only Jira administrators, runs the call with the user's own permissions, and sends nothing on its own initiative. The AI client is chosen and authorised by your organisation and its user, so its provider's terms and retention apply to the answers it receives; it is not a Radrly sub-processor. Admins can turn the tools off per site at any time, and users can revoke the client's access in their Atlassian account.
Each tool call is recorded in the AccessRadar audit log (account ID of the caller, action and number of rows) and follows the audit retention setting.
Permissions (scopes)
Jira asks you to approve the following kinds of scopes when the app is installed. No write or manage scopes are requested.
| Scope family | Why AccessRadar needs it |
|---|---|
Space and catalogue read:* | List spaces and related metadata so schemes and roles can be mapped. |
read:permission-scheme:jira, read:permission:jira | Read permission schemes and grants. |
read:project-role:jira, read:group:jira, read:user:jira, read:application-role:jira | Resolve who holds access and through which path. |
report:personal-data | Report stored Atlassian account IDs to Atlassian's personal data reporting API daily, and replace the stored display name with a pseudonym when Atlassian reports an account closed. |
read:filter:jira, read:dashboard:jira, read:jql:jira | List filters and dashboards and who they are shared with, so the app can find items shared with anyone or all signed-in users. read:jql:jira is required by Jira in the filter search response; the JQL is not shown or stored. |
read:issue-security-scheme:jira, read:issue-security-level:jira | Show issue security schemes, levels, who is granted each level, and which spaces use the scheme. Work items are not read. |
read:audit-log:jira | Show permission, group and role changes of the last 7 days, with before and after values (Advanced). Jira exposes the audit log on paid plans only. |
See Security for the full scope discussion, including app-user group membership and optional impersonation for collection fallback.
Retention and deletion
- While installed: in the Standard edition, snapshots older than 90 days are deleted automatically. In the Advanced edition, you choose how long to keep history in Settings, configurable from 30 to 3650 days. The audit log and in-app notices follow the same retention window as snapshots. Signed reviews, exceptions, campaigns and assignments are kept while the app is installed so evidence remains available. A snapshot that a review is based on is kept with that review.
- In-app deletion: admins can delete draft reviews in the app. Signed reviews cannot be deleted in the app. For all other data, uninstall removes the app's data under Atlassian's Forge retention rules.
- After uninstall: the Forge platform soft-deletes the app's data and destroys it at the end of the retention period described in Atlassian's SOC 2 report. A reinstall within 21 days can be relinked to the old data. This period is set by Atlassian. Radrly itself keeps no copy.
- Personal data reporting: once a day the app reports stored Atlassian account IDs to Atlassian's personal data reporting API. An account Atlassian reports as closed is not reported again. For a closed account, AccessRadar replaces the stored display name in every stored version with a pseudonym (
Closed accountplus a short hash) and re-applies it daily, so a later snapshot cannot bring the name back. The Atlassian account ID is kept as a pseudonymous key in snapshots, review items, sign-offs and the audit log, so that signed evidence stays verifiable. This is pseudonymisation, not anonymisation. Free-text notes and justifications are not changed. In-app alert notices can quote a display name and are deleted when they pass the retention window.
Logs
The app writes operational status and error messages to the Forge developer console. Logs stay in Atlassian's environment and are not sent to Radrly.
GDPR
Roles. For the personal data in AccessRadar (account IDs, display names, group membership and role assignments reflected in snapshots, and review decisions, exceptions, campaign assignments, notices and audit-log entries), the Customer is the controller and Radrly is a processor acting on the Customer's documented instructions. A Data Processing Agreement is available.
Legal basis. The Customer decides the legal basis for reviewing access of its personnel (for example legitimate interests or compliance with legal obligations). Radrly does not decide the purposes of the processing.
Your rights. If you are a person whose data is in a Customer's AccessRadar, you have rights of access, rectification, erasure, restriction, portability and objection under the GDPR. Please contact the organisation that runs your Jira site first. We will support the Customer in answering such requests. You also have the right to complain to a supervisory authority (in Poland: the President of the Personal Data Protection Office, UODO).
Security. See Security.
International transfers
Radrly does not transfer data itself. Data is hosted by Atlassian in the location set by the Customer's data residency choice, and Atlassian's own transfer mechanisms apply.
CCPA
Radrly does not sell personal information. Radrly is an EU company; any California consumer personal information processed through the app is handled only within Atlassian's platform under Atlassian's terms.
This website
This section covers the AccessRadar website (accessradar.radrly.com), not the AccessRadar app. The app runs on Atlassian with zero egress, and nothing below changes that.
Server logs. Our web server keeps standard access logs (IP address, time, requested page) for security and operations. The logs are rotated daily and kept for 14 days. Legal basis: our legitimate interest in running a secure site (Art. 6(1)(f) GDPR).
Videos. The explainer video loads from YouTube (youtube-nocookie.com) only when you press play. From then on Google processes data under its own privacy policy. This does not depend on the cookie choice below.
Cookies and Google tags
Before you choose, no Google tag loads and no data goes to Google. The site shows a banner with Accept and Reject. Your choice is stored in a first-party cookie, radrly_consent, on radrly.com sites for 180 days. It holds only the choice and its date. This cookie is strictly necessary.
If you accept, the site loads the Google tag (gtag.js) for two purposes:
- Google Analytics measures visits, pages, traffic sources and clicks on the Atlassian Marketplace link. Cookies:
_gaand_ga_*, kept for 6 months. Google Signals and ad personalisation are off. - Google Ads conversion tracking measures which of our ads lead to a click to the AccessRadar listing on Atlassian Marketplace. Cookies:
_gcl_auand, if you came from one of our ads,_gcl_aw, kept for 90 days. Google may also set cookies on its own domains, as described in Google's policies. We do not use remarketing or personalised ads.
Data processed: online identifiers in these cookies, IP address (used by Google to send the response and for coarse location; Google Analytics does not store IP addresses), browser and device data, pages viewed, the referring page and campaign parameters. Reports are aggregated. Google Analytics event data is kept for 14 months.
Provider. Google Ireland Limited, Gordon House, Barrow Street, Dublin 4, Ireland, acting as our processor for Google Analytics and as an independent controller for Google Ads measurement. Data may be transferred to Google LLC in the United States, which is certified under the EU-U.S. Data Privacy Framework.
Legal basis. Your consent (Art. 6(1)(a) GDPR, and Art. 399 of the Polish Electronic Communications Law for storing and reading cookies).
Withdraw or change your choice at any time: click Cookie settings at the bottom of any page and choose Reject. The Google tag stops, and the site deletes the _ga and _gcl cookies it can reach. Withdrawing does not affect processing before it. You can also delete cookies in your browser.
Links to Atlassian Marketplace. When you click a link to the AccessRadar listing, the site adds the campaign parameters (utm_*) from the address you arrived with, so we can see which campaign led to an install. With your consent it also adds the Google click ID. Atlassian Marketplace is run by Atlassian under its own privacy policy.
Your rights are listed above (access, rectification, erasure, restriction, objection, portability and a complaint to the supervisory authority). Contact support@radrly.com.
Changes
We will update this page when the app's data handling changes and show the date of the latest update at the top. For material changes we will notify customers, for example through Atlassian Marketplace.
- Version 1.6 (2026-11-12): AI pre-sort for reviewers (app 4.0.0, Advanced): optional, Jira-admin-triggered Atlassian Forge LLMs; minimised input; suggestions not stored and not used for training; advice only.
- Version 1.5 (2026-11-12): Sharing and security (app 3.0.0) reads filter and dashboard sharing, issue security levels and (Advanced) Jira audit-log summaries live. This data is not stored. Six read-only scopes were added. No write scopes and no egress.
- Version 1.4 (2026-11-11): AI and Rovo: AccessRadar's read-only Rovo actions are also available as Rovo and MCP tools; external AI clients only when your admin enables them.
Contact
Privacy questions and data requests: support@radrly.com. Security contact: support@radrly.com. We have not appointed a Data Protection Officer; privacy questions go to the contact above.