Privacy Policy: miniOrange Secure MCP for Atlassian (Jira & Confluence)

miniOrange is committed to protecting the privacy and security of information processed through its products and services. This Privacy Policy explains how Xecurify Inc. ("Xecurify", "miniOrange", "we", "our", or "us") collects, processes, uses, stores, and shares information in connection with miniOrange Secure MCP for Atlassian (Jira & Confluence) (the "Service").

The Service allows authorized AI assistants and other Model Context Protocol (MCP) compatible clients to interact with Jira and Confluence in Atlassian Cloud, subject to the permissions and access rules configured by the Atlassian administrator.

This Policy applies only to the Service.

1. Overview

Secure MCP for Atlassian provides a controlled connection between an external AI assistant, such as Claude or ChatGPT, and your Atlassian Cloud site. Instead of connecting the AI assistant directly to Atlassian, your administrator connects it using a URL provided by the Service.

Before a request is sent to Atlassian, the Service checks it against access rules defined by your administrator and records the request in an activity log.

The Service has two components. They are described separately because each component handles and stores information differently:

• The Atlassian Forge application. This application runs inside your Atlassian Cloud tenant on infrastructure operated by Atlassian. Your access rules, your users' Atlassian tokens, and your activity log are stored in Atlassian-hosted storage created for your installation. miniOrange does not operate a separate server that stores this data and cannot read it.
• The connection router. This routing service is operated by miniOrange at atlasmcp.miniorange.in. It forwards requests to your installation and adjusts sign-in information so authentication can complete on a domain controlled by miniOrange. It does not keep a database, registry, or credentials. Your installation address is carried in the request URL instead of being stored in a lookup table.

Your Atlassian administrator decides which Atlassian groups, operations, Jira projects, and Confluence spaces are available to connected AI assistants.

2. Information We Process

2.1 Atlassian Site Connection and Configuration Data

The following information is stored inside your Atlassian Cloud tenant:

• Access rules. Rules created by your administrator may include the AI assistant identifier, Atlassian group names, operation names, Jira project keys, Confluence space keys, whether a request is allowed or denied, and the default action when no rule matches.
• Atlassian API credential. Your administrator supplies this credential so the Service can list the operations available on your site. It is stored as an encrypted secret in Atlassian Forge. It is never shown again in the interface, written to logs, or made accessible to miniOrange. Your administrator can clear it at any time.
• Site information. Your site cloud identifier, site name, and host name are used to label activity records and route requests to the correct site.

2.2 Atlassian Data Processed Through the Service

Using the permissions approved during installation, the Service reads the following information from your Atlassian site so it can apply access rules and label activity records:

• User profile information. Atlassian account identifier, display name, email address, avatar URL, and account type.
• Group memberships. The Atlassian groups a user belongs to. This information is used to apply group-based access rules.
• Jira project and Confluence space identifiers. These identifiers are checked live for each request so project-specific and space-specific rules can be applied.

Jira issue content and Confluence page content pass through the Service when an AI assistant requests them, but the Service does not store that content. See section 2.3.

2.3 MCP Tool Requests and Responses

• Allowed and denied requests. If an access rule allows a request, the Service sends it to Atlassian's Rovo MCP server using the signed-in user's own Atlassian token. If an access rule denies the request, it is not sent to Atlassian.
• Response content is not stored. The Service records only the response size in bytes and the time the request took. The response body, including issue fields, page content, and search results, is returned to the AI assistant and is not retained by the Service.
• Request values are redacted before storage. The activity log records which argument names were present, but replaces each value with [redacted], except for a fixed list of low-risk identifiers. These identifiers are stored as provided and limited to 80 characters: cloudId, issueKey, issueIdOrKey, projectKey, projectId, spaceKey, pageId, parentId, boardId, and sprintId. Free-text values such as issue summaries, comment bodies, and page content are not written to the activity log.

Each activity record may contain the following information:

• the time of the request;
• the acting user's Atlassian account identifier, display name, email address, avatar URL, and account type;
• the authentication method and the user's group memberships;
• the AI assistant identifier and the operation requested;
• the redacted request arguments described above;
• the access decision and the rule that produced it;
• the outcome status and HTTP status code;
• the response size and duration;
• the Jira project or Confluence space resolved for the request, and whether it could be resolved;
• the reason a request was blocked, when applicable;
• the Atlassian product;
• your site's cloud identifier, site name, and host name;
• a session identifier; and
• a one-line summary of the request.

Activity records are stored in a Forge SQL database created only for your installation inside your Atlassian Cloud tenant.

2.4 Authentication and Authorization Data

• Your Atlassian tokens. When a user signs in, the access token, refresh token, and expiry information issued by Atlassian are stored as encrypted Forge secrets linked to that user's Atlassian account identifier. This allows each request to run under the user's own account and permissions. A separate non-secret list of account identifiers is kept so an administrator, or the uninstall process, can find and delete stored tokens.
• Tokens issued to your AI assistant. The Service operates an OAuth 2.1 authorization server. Access tokens expire after 1 hour and refresh tokens expire after 30 days. These tokens are stored in your installation's storage, expire automatically, and are removed when they expire or when an administrator disconnects the user.
• AI client registration. When an AI assistant connects for the first time, it registers automatically. The Service stores the client identifier, the redirect addresses supplied by the client, the client name reported by the client, and the registration time.
• Temporary sign-in data. Short-lived authorization codes, PKCE code challenges, and sign-in request state are kept only for the duration of the sign-in process and are then discarded.

The Service does not collect, transmit, or store passwords.

2.5 Operational and Security Data

• Application logs inside your Atlassian tenant. The Service writes one structured log entry per request. This entry contains no request arguments or personal data. Additional diagnostic logging for sign-in activity and calls to Atlassian is disabled by default. When enabled for troubleshooting, it does not include tokens, request arguments, or personal data.
• Router access logs. The connection router keeps a minimal access log containing the HTTP method, the requested path with your installation identifier removed, the status code, and the request duration. It does not log request bodies, headers, or tokens.
• Rate limit counters. Request counters are held in memory for each installation and expire automatically. They are not written to disk.
• Web server logs. The web server in front of the router keeps standard access logs that may include the client IP address. These logs are used for security monitoring and abuse prevention.

2.6 Information We Do Not Collect

The Service does not collect:

• your AI assistant conversation history or prompts;
• the content of Atlassian records returned to your AI assistant;
• payment card data;
• health information;
• government-issued identifiers;
• biometric data; or
• passwords or account credentials for your Atlassian account.

3. How We Use Information

The information described in section 2 is used only to operate and protect the Service:

• to authenticate users and carry out requests under each user's own Atlassian permissions;
• to check each request against the access rules defined by your administrator;
• to keep an activity log so your administrator can review what was requested, allowed, and blocked;
• to route each request to the correct Atlassian installation; and
• to protect the Service against abuse and troubleshoot service issues.

miniOrange does not sell Personal Data processed through Secure MCP for Atlassian.

miniOrange does not use Atlassian data processed through the Service to train general-purpose AI models.

miniOrange does not use your activity log, access rules, or Atlassian content for advertising, profiling, or any purpose other than operating the Service. This data is stored inside your Atlassian Cloud tenant, and miniOrange cannot access it.

4. Access Control and Customer Responsibility

As the Atlassian site administrator, you control what connected AI assistants can access. You are responsible for:

• Defining access rules. Set the rules that allow or deny requests, including the default action when no rule matches.
• Restricting sensitive content. Explicitly restrict Jira projects and Confluence spaces that should not be available to AI assistants.
• Choosing AI assistants. Decide which AI assistants may connect and share the connection URL only with approved users or clients.
• Protecting the connection URL. Treat the URL as sensitive because it contains a credential that identifies your installation. Anyone with the URL can reach your installation's sign-in endpoint.
• Limiting API credential access. Provide an Atlassian API credential with only the permissions needed for the Service and clear it when it is no longer required.
• Reviewing activity. Review the activity log and disconnect users who should no longer have access.
• Informing users. Where required by applicable law, inform your users that AI assistant activity involving your Atlassian site is recorded.

5. Sharing and Recipients

In normal operation, the Service shares or routes data only with the parties needed to provide the Service:

• Atlassian. Permitted requests are sent to Atlassian's Rovo MCP server so they can be carried out in your Atlassian Cloud site. The Service also reads user, group, project, and space information from Atlassian APIs to apply your access rules. Atlassian already holds this data on your behalf, and its handling is governed by your existing agreement with Atlassian.
• Your AI assistant provider. The result of a permitted request is returned to the AI assistant connected by your administrator and therefore to the provider operating that assistant. This happens only for requests allowed by your access rules. The provider's handling of the data is governed by your agreement with that provider. See section 7.
• miniOrange hosting infrastructure. We use third-party cloud providers to host and operate the Service (cloud hosting, a managed database, and a managed cache). They process data only to provide hosting to us under contract.

A current list of sub-processors is available on request at info@xecurify.com.

miniOrange does not sell, rent, or share your data with advertisers, data brokers, or analytics providers. miniOrange does not disclose information to other third parties except when required by law or valid legal process.

6. Data Retention

Data	Retention Period
Access rules	Until changed or deleted by your administrator, or until the app is uninstalled
Atlassian API credential	Until cleared by your administrator, or until the app is uninstalled
Users' Atlassian tokens	Until the token expires, the user is disconnected by an administrator, or the app is uninstalled
Access tokens issued to the AI assistant	1 hour
Refresh tokens issued to the AI assistant	30 days
Activity log records	Until cleared by your administrator, or until the app is uninstalled
Cached site name and host name	Up to 24 hours
Cached site cloud identifier	Up to 24 hours
Cached Atlassian group list	Approximately 1 hour
Cached operation list	Until an administrator selects "Refresh tools"
Jira project and Confluence space lookups	Not cached. They are checked live for every request.
Router rate limit counters	Held in memory only and deleted automatically when they expire

7. Third-Party AI Platform Data

The Service does not receive, store, or have access to your conversations with your AI assistant (ChatGPT, Claude). It receives only the specific operation requested by the assistant and the arguments sent with that operation at the time the request is made.

How your AI assistant provider handles Atlassian data returned to it, including whether the data is retained, logged, or used for model training, is determined by that provider and your agreement with them, not by miniOrange. Administrators should review the provider's terms before allowing a connection.

8. Data Protection Roles

• You are the data controller for data stored in your Atlassian Cloud tenant. This includes Atlassian content, user information, access rules, and activity log records. As the data controller, you determine how this data is used. miniOrange cannot access this data and does not act as its processor.
• miniOrange acts as a data processor for traffic passing through the connection router. This role is limited to forwarding requests on your behalf and carrying out the operational logging described in section 2.5.

9. Your Rights and Choices

Depending on the law that applies to you, you may have the right to access, correct, delete, restrict, or object to the processing of your Personal Data, and to request a copy of it.

Because the data described in sections 2.1 to 2.4 is stored inside your Atlassian Cloud tenant, your Atlassian administrator can use the app to manage that data:

• Disconnect a user. Deletes that user's stored Atlassian token. The user must sign in again before using the connection.
• Clear the activity log. Deletes all activity records for the installation.
• Clear the Atlassian API credential. Removes the stored credential.
• Uninstall the app. Removes the data stored for the installation.

10. Security

• User permissions are preserved. Requests use each user's own Atlassian OAuth token, so the Service cannot access data beyond that user's existing Atlassian permissions. No shared or service account is used for user requests. If a user token is missing or expired, the request fails instead of using another credential.
• Access rules are checked before requests are sent. The Service applies access rules on the server before forwarding a request to Atlassian. Changing instructions given to the AI assistant cannot bypass these rules.
• Credentials are stored securely. The Atlassian API credential and each user's Atlassian tokens are stored as encrypted Forge secrets. They are not returned to the interface or written to logs.
• Request values are redacted. Request arguments are redacted before they are written to the activity log, as described in section 2.3.
• The router limits where requests can be sent. The connection router sends requests only to Atlassian Forge web-trigger addresses and rejects other destinations. It also applies rate limits for each installation and across the Service.
• Installation identifiers are removed from router logs. The router removes the part of the URL that identifies your installation so the credential contained in the URL is not written to the access log.
• Administrative features are restricted. Server-side checks limit administrative functions to users with Jira administrator permission.

No method of transmission or storage is completely secure, and we cannot guarantee absolute security.

11. International Transfers

miniOrange and its service providers may process information in countries other than the country where the customer or user is located.

When Personal Data is transferred internationally, we use appropriate safeguards as required by applicable data protection laws.

Where required, these safeguards may include contractual protections or other legally recognized transfer mechanisms.

12. Children's Privacy

The Service is not directed to children under 13, and we do not knowingly collect personal data from them.

If you believe Personal Data relating to a child has been provided directly to miniOrange inappropriately, please contact us so we can review the issue and, where appropriate, delete the information.

13. Changes to This Policy

We may update this Policy to reflect changes to the Service or legal requirements. The "Last Updated" date at the beginning of this Policy shows when it was most recently revised. If a change materially affects how Personal Data is processed, we will provide notice through the product listing or the app interface.

Contact Us

If you have questions, concerns, or requests about this Privacy Policy or miniOrange privacy practices, you can contact us using the details below:

Email: info@xecurify.com

Phone: +1 978 658 9387

Form: Contact Us / Visit miniOrange