Skip to content
Agent Orchestration
Skill

/jira-execute

Implement a Jira issue, open a pull request, and report the work on the issue. Insert into the prompt of a Jira column rule that does the building.

BOOST
From plugin
studio
41224 skills
Install
$ npx -y skills add decocms/studio --skill jira-execute --agent claude-code

How it fires

How this skill gets triggered: by you, by Claude, or both.

  • Fires itselfAuto-invocation. Claude auto-loads it when your prompt matches the work.Auto-invocation is when the right skill fires by itself at the right moment, driven by a FLOW.md router and a hook, instead of you invoking it by name. It is the difference between a skill being installed and a skill actually getting used.Read the full definition →
  • You can call itInvoke it directly when you want it.
  • Slash command/jira-execute

Context preview

The summary Claude sees to decide when to auto-load this skill.

Implement a Jira issue, open a pull request, and report the work on the issue. Insert into the prompt of a Jira column rule that does the building.

SKILL.md

jira-execute.SKILL.md
name: jira-execute
description: Implement a Jira issue, open a pull request, and report the work on the issue. Insert into the prompt of a Jira column rule that does the building.
disable-model-invocation: true

jira-execute — implement the issue and hand over

Insert this into the prompt of the Jira column your team builds in. It is the first of two runs: this one implements and hands over, and a later column reviews what it left (`jira-review`).

It is ordinary text once inserted — read it, cut what does not apply to your team, add what does.

---

You are implementing a Jira issue. The issue itself is above.

Build it

  • Make the change, commit it, push the branch, and open a pull request.
  • Change only what the issue needs. Don't refactor around it.
  • The change must be REACHABLE from the surface the issue names: edit the

component that route actually renders, not one that merely looks like the right place. A change nothing imports is the most common reason work comes back rejected.

  • Open the pull request from the branch you were given. Don't move the work to

a differently-named one.

Verify it on the deploy preview

Nobody else will. Do not hand over on a green test suite.

  • After you push, the deploy bot comments the preview URL on the pull request —

poll `gh pr view <n> --json comments` until it appears (it takes a couple of minutes), then open that URL. A first request may wake a hibernating build; retry until you get real HTML.

  • Exercise the actual behaviour there and MEASURE it. Read the value back from

the page rather than inferring it from a name or a spec — a spec that disagrees with production is a finding worth reporting, and it is the kind of thing only this step catches.

  • Capture evidence with `qa-screenshot <url> <path>.png [--mobile]`, on desktop

AND mobile, and `Read` each file. A screenshot you never opened is not verification, and mobile is not desktop resized.

Report on the ISSUE

There is no card for anyone to read. A final message in this run is not a report — the issue is the only place your work becomes visible.

  • `JIRA_REMOTE_LINK_ADD` puts the pull request on the issue as a link

(`key: "pull-request"`), and the deploy preview as another (`key: "preview"`). Do both — a URL buried in a comment is not something a person clicks.

  • `JIRA_COMMENT_ADD` posts your report (markdown, tables included). Say what you

changed, what you MEASURED on the preview, and what you deliberately left alone. Name anything the issue asked for that you did not do.

  • To show evidence, write each screenshot to `/app/org/output/<name>.png` and

reference it in that comment as `![what it shows](/app/org/output/<name>.png)` — it is uploaded to the issue and rendered inline. A before/after table of two images reads best.

  • `JIRA_ISSUE_GET` re-reads the issue. Worth doing before you report: a person

may have edited the card while you worked.

  • `JIRA_ATTACHMENT_DOWNLOAD` fetches an attachment into the pod by its id, when

the card carries a mockup or a log you need.

  • `JIRA_ISSUE_TRANSITION` hands the issue over — LAST, after the links and the

comment are on the card. Move it to the column where this work gets REVIEWED and **no further**: that is where the next run picks it up. Landing it in a QA or customer-facing column skips the review entirely and hands unverified work to a person. Then **re-read the issue** and confirm it actually landed there — a transition's advertised destination is not always where it puts the card, and the failure is silent.

If the issue turns out to need no code change, do not open a pull request: say so in the comment, with what you checked, and move the issue on.

Read more
Ships withstudio

Open-source control plane for your AI agents. Connect tools, hire agents, track every token and dollar

Get the whole plugin

Other skills on studio.