jira-writinglisted
Install: claude install-skill atomgunlk/skills
# Writing Jira issues (Atlassian MCP)
## Overview
Two facts drive everything here:
1. `editJiraIssue` **replaces the entire description** โ there is no append/patch. Always `getJiraIssue` first (someone may have edited it) and send the whole body back.
2. `contentFormat: "markdown"` authors prose, headings, bullets, tables, and code blocks fine โ but it **cannot express ADF-only nodes**: smartlink chips (`inlineCard`), mentions, status lozenges, dates, emoji, media. Re-saving a body that contains any of them as markdown silently destroys them.
## Creating a new issue
- `createJiraIssue` with `contentFormat: "markdown"` for plain content. Issue type names vary per project ("Story" vs localized or renamed types) โ when unsure, confirm via `getJiraProjectIssueTypesMetadata` / `getJiraIssueTypeMetaWithFields` (also reveals required custom fields) instead of guessing `issueTypeName` and letting the create call fail.
- Description shape: **Context โ Acceptance criteria (AC table, see below) โ Out of scope**.
- Chip needed in a brand-new issue โ author the whole doc as ADF (`contentFormat: "adf"`) from the start.
## Acceptance criteria โ AC table (house style)
Acceptance criteria in a description ARE a markdown table โ **one row per AC, five columns**. Not "shall" clauses, not free-form bullets. If the source arrives as EARS, prose, or Gherkin `Scenario:` blocks, convert it into rows (a Gherkin `Scenario:` title maps straight onto the `Scenario` column).
| AC No. | Scenario