Katabarwa Labs
← All Atlassian apps

Jira · Agent Dispatch for Jira

Send a Jira issue to Claude Code or Codex in your own CI, and get the pull request back on the ticket

One button on the issue fires your GitHub Actions or GitLab CI. The agent runs with your API key on your runners, and the PR link and status come back to the issue.

Runs on Atlassian Forge inside your own tenant. No vendor servers: its only network calls go to api.github.com and gitlab.com, the CI you already own.

The gap

The assign-to-agent connectors on the Marketplace run the agent on the vendor's infrastructure, speak GitHub only, ignore the base branch, and make you type the repository into a chat box every time. Engineering teams that want ticket-to-PR automation end up choosing between a vendor holding their code and keys, or no automation at all. GitLab shops get no choice.

What it does
The issue panel: one click sends the work item to your CI, and the run status and pull request link come back on the issue.
The issue panel: one click sends the work item to your CI, and the run status and pull request link come back on the issue.
Project settings: repository, base branch, provider and agent, configured once, with a live check of the repository setup.
Project settings: repository, base branch, provider and agent, configured once, with a live check of the repository setup.
The result: a pull request opened by your own GitHub Actions, on your base branch, ready to review.
The result: a pull request opened by your own GitHub Actions, on your base branch, ready to review.
Honest about its limits
Security & permissions

Built on Atlassian Forge: the app runs on Atlassian's own serverless platform inside your tenant and stores its data in Forge storage. Its network calls are declared in the manifest and limited by Forge to api.github.com and gitlab.com; no other destination is reachable. What leaves your tenant is the summary and description of an issue someone explicitly dispatched, sent to the CI pipeline you configured for that project. Nothing is sent anywhere else, and we host no servers. Every scope it requests, and why:

read:jira-work the summary, description and key of the issue being dispatched, and the issue panel context
write:jira-work the one result comment written back onto the originating issue
storage:app project mappings, governance settings, run records and the audit log; CI credentials in Forge secret storage, write-only
The native feature it builds on

Atlassian's own documentation for what Jira does out of the box, so you can see exactly where the gap is:

FAQ

Does any data leave my instance?

Only to systems you own. The app runs on Atlassian Forge inside your tenant and calls api.github.com and gitlab.com, nothing else. What leaves is the summary and description of an issue someone explicitly dispatched, sent to the CI pipeline you configured for that project. Nothing is sent anywhere else, and we host no servers.

How is it billed?

Through the Atlassian Marketplace, on your existing Atlassian bill, priced per user like any Cloud app. The listing has a free evaluation and the exact calculator for your tier.

Where do I get support?

Email support@llmgraph.ai or open an issue from the documentation page. You talk directly to the people who build the app.

Install Agent Dispatch for Jira in about a minute.

Free evaluation from the Marketplace listing. If it does not fit your workflow, tell us what would and we will build it.