Read-only Jira scope
read:jira-work permits retrieval of required Jira data through Jira user context. The app does not request Jira write permissions.
Security and trust
ReleaseProof uses a read-only Jira permission model, deterministic rules, minimized frontend results, and fail-closed processing.
Security posture
read:jira-work permits retrieval of required Jira data through Jira user context. The app does not request Jira write permissions.
storage:app stores ReleaseProof project configuration within the Forge app context.
The readiness engine uses code-based deterministic rules. It does not send Jira data to an LLM.
Full descriptions and acceptance text sources are excluded from the public analysis value.
Incomplete pagination, malformed pages, or unsafe issue mappings abort the analysis.
ReleaseProof presents evidence. It neither changes Jira nor makes the release decision.
Data handling
Assurance boundaries
Hosting and processing boundary
The Atlassian-hosted application architecture runs release analysis inside Atlassian Forge. The public ReleaseProof website is hosted separately on Vercel. Jira release-analysis data is not processed by the marketing website.
The current Jira analysis path does not use an external product database, generative AI provider, or marketing-site service. Vercel hosts only this public website. Lead delivery is a separate, deployment-configured service and must be documented before launch.
Report a suspected security issue through the contact route. Do not include credentials, production Jira content, or personal data in the initial message.
Contact securityUse the contact form to request architecture, permission, and data-flow details relevant to your evaluation.