Skip to content

Salable/atlassian-cursor-skill

Repository files navigation

Cursor Atlassian context

Cursor augmentation for Atlassian (Jira + Confluence): custom commands and a skill that load Jira/Confluence data into context before code generation. Uses acli for Jira and a small Confluence script.

Prerequisites

  • acli (Jira)
  • curl and jq (for the Confluence script)
  • Cursor

Setup

  1. Credentials: Copy .env.example to .env and set:

  2. Jira (acli): One-time auth from the repo root:

    ./.cursor/scripts/fetch-atlassian.sh --acli-auth
  3. Use in Cursor: When you want Atlassian data in context, run one of the commands below and include your request in the same message.


How the commands work

Each command tells the Cursor agent to fetch Atlassian data first, use that output as context, and only then answer or implement your request. The agent will run the right CLI/script (acli for Jira, fetch-atlassian.sh for Confluence), paste the result into context, and then code or answer from that.

  • /jira — Use when your task depends on Jira issues. Give an issue key, JQL, or filter ID plus what you want done (e.g. implement, summarize, plan).
  • /confluence — Use when your task depends on Confluence pages. Give a page ID or a search (space key, text, CQL) plus what you want done (e.g. implement from spec, write a guide, compare with code).
  • /atlassian-context — Same idea for either Jira or Confluence (or both); use when you’re not sure which or want to combine them.

If you don’t give a key, JQL, page ID, or search, the agent will ask for it.


Using the commands

Type the command and your request in one message. Examples:

Jira — single ticket

  • /jira PROJ-123 Implement the acceptance criteria from this ticket
  • /jira PROJ-456 Add a test that covers the edge case in the description
  • /jira PROJ-789 Summarise this ticket and suggest a branch name

Jira — search or “my work”

  • /jira assignee = currentUser() and status = "In Progress" List these and suggest what to tackle first
  • /jira project = MYPROJ and sprint in openSprints() Show the sprint backlog
  • /jira filter 12345 Summarise the top 5 issues and suggest an order to work on them

Confluence — page by ID

  • /confluence 12345678 Implement the API described on this page
  • /confluence 12345678 Turn this spec into a checklist in the repo README

Confluence — search (build local context from docs)

  • /confluence Search for pages about "API authentication" and summarise how we do auth
  • /confluence text ~ "metered billing" --limit 5 Write a short guide from these pages
  • /confluence space=DEV and type=page order by lastmodified desc --limit 10 Create an onboarding doc from the 10 most recent DEV pages
  • /confluence space=SA and type=page order by lastmodified desc Pull in the latest product docs and suggest where the codebase might be out of date

Combined flow (ticket + docs)

  • /atlassian-context Get Jira PROJ-100 and Confluence page 9999999, then implement the ticket using the spec on the Confluence page

Example workflows

Building a feature from a Jira ticket

  1. Open the ticket in your head (e.g. PROJ-123).
  2. In Cursor Agent: /jira PROJ-123 Implement the acceptance criteria; create the branch and code from the ticket.
  3. The agent loads the issue (title, description, AC), then creates the branch and implements.

Building local context from Confluence before coding

  1. You want to align code with internal docs (e.g. “billing” or “auth”).
  2. In Cursor Agent: /confluence text ~ "billing API" --limit 5 Summarise what these pages say our billing API should do, then check our code in src/billing and list gaps.
  3. The agent runs the Confluence search, gets the content, then compares with the codebase.

Checking Jira details while working

  1. You’re mid-task and need the latest ticket details or comments.
  2. In Cursor Agent: /jira PROJ-456 Show the latest comments and description; update the test file tests/proj456_test.go to match the new edge cases.
  3. The agent fetches the issue (and comments), then updates the test file from that context.

Onboarding or “state of the docs”

  1. You want a snapshot of recent Confluence content (e.g. space KEY).
  2. In Cursor Agent: /confluence space=KEY and type=page order by lastmodified desc --limit 10 Summarise these pages as a one-page onboarding or “recent docs” note.
  3. The agent fetches the pages and produces the summary.

Docs

About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors

Languages