Jira

Anthara connects to Jira Cloud with each person’s own connection. Workflows use that connection to read issues and to let an agent update them.

Jira cannot start a workflow. There is no Jira trigger. To act on Jira issues, run a workflow by hand or on a schedule, and have its first step read the issues.

The connection

A connection is your own sign-in to Jira through an OAuth app. To create one, see Connections.

Anthara asks Atlassian for these scopes:

Scope	Why
`read:jira-work`	Read issues and search with JQL.
`write:jira-work`	Create and edit issues, and add comments.
`manage:jira-project`	Manage project settings such as components.
`read:jira-user`	Read user details.
`offline_access`	Stay connected without asking you to sign in again.

Read issues in a workflow

A read_data step with the jira provider runs a JQL query. JQL is Jira’s query language.

- name: fetch-stories
  type: read_data
  provider: jira
  connection: Jira
  config:
    query: >
      project = DEMO AND status = "To Do"
      ORDER BY priority DESC

Key	Required	Meaning
`connection`	Yes	The name of the Jira app on the Connections page.
`config.query`	Yes	The JQL query.
`config.fields`	No	The issue fields to return.

When you leave out config.fields, each issue comes back with summary, description, status, priority, issuetype, assignee, reporter, labels, components, and comment.

Two limits to know:

One page of results. The step makes one search request and does not fetch further pages. Write the JQL so that the issues you need fit in the first page.
The first Jira site. If your Atlassian account can reach more than one Jira site, the step uses the first one Atlassian returns.

The step’s output is a list of issues. The next step then runs once per issue. See Fan-out.

Let an agent update Jira

An ai_task step can be given the Jira connection. The agent receives an access token and calls the Jira REST API itself.

- name: comment-on-story
  type: ai_task
  connections: [Jira]
  config:
    llm_config:
      system_prompt: |
        You work inside a container. Call the Jira REST API with curl and
        post the comment yourself.
    prompt: |
      You receive one Jira story as JSON:
      {{steps.fetch-stories.output}}

      Add one comment to the story that says whether it is ready to build.
      Use the token in the JIRA_ACCESS_TOKEN environment variable. Never print it.

A connection named Jira reaches the agent as the environment variable JIRA_ACCESS_TOKEN. The name is turned into upper case, and every character that is not a letter or digit becomes an underscore.

The token works against api.atlassian.com. The agent first asks https://api.atlassian.com/oauth/token/accessible-resources for the site’s cloud ID, and then calls https://api.atlassian.com/ex/jira/<cloud ID>/rest/api/....

Set a system_prompt on a step like this one. Without it, the agent is told not to run commands, and the step can finish without changing anything.

Whose connection is used

How the run starts	Connection used
A person selects Run	That person’s Jira connection
A cron schedule	The workflow creator’s Jira connection

A complete example

The plugin ships a working example that reads Jira stories and comments on each one. In Claude Code, run /anthara:workflow-builder to start from it.