Overview
Jira records what a team has agreed to do. AI is useful around it: turning a meeting into tickets, summarising a sprint, finding duplicates, writing the status update nobody wants to write.
Kroy connects to Jira once. Authorised AI tools reach Jira through Kroy, which knows who each request is for and limits it to the projects that person works on.
Permissions
Each Jira action is a Kroy capability. Kroy decides every request against the person, the agent, the delegation between them and the organisation’s policies.
By default, AI may search, read, create issues, comment and update fields. Moving an issue through the workflow — for example to Done or to Released — requires approval, because transitions often trigger automation elsewhere. Organisations can relax or tighten this per project.
Jira and Kroy State
Jira holds the tickets. Kroy State can hold the working context around them: the decisions behind a ticket, the notes an agent left for the next agent, the link between a Jira issue and the GitHub branch that fixes it. Any authorised AI can pick up where another left off.
Industries
Technology companies use Jira for engineering work. Consultancies often use it to run client projects, where a partner may want AI to prepare weekly status from Jira without exposing other clients’ projects.
Security
Kroy holds the Atlassian connection. AI clients never see the tokens. Every request is recorded with the person, agent, project and issue involved, and revoking an AI’s Kroy access removes its Jira access immediately.