About Jira Delivery
Jira Delivery is a Jira Cloud MCP server and agent skill for coding agents. It prepares ticket briefs, checks local branch changes against acceptance criteria, and creates QA handoffs; it is read-only by default, with approved, previewed write-back available as an option.
Key features
- 01Prepare ticket work briefs
- 02Check code changes against acceptance criteria
- 03Generate traceability matrices
- 04Create QA handoffs, release notes, and test plans
- 05Search Jira and Confluence content
- 06Write back to tickets with approval
Use cases
- 01Identify blockers and next steps before starting a ticket
- 02Check a branch against acceptance criteria before a PR
- 03Prepare QA handoffs and test plans
- 04Search project documentation for ticket context
MCP support
The MCP server connects to Jira Cloud and can search Confluence, exposing tools for ticket preparation, change checks, and handoffs to clients. It is read-only by default; write operations can be enabled and require preview and approval.
Configuration
Local server config is the same for every client. Copy it into your client’s MCP config file.
npx -y jira-delivery-mcpJIRA_BASE_URLYour Jira Cloud site, e.g. https://your-domain.atlassian.netJIRA_EMAILEmail of the Atlassian account the API token belongs toJIRA_API_TOKENSecretAtlassian API tokenJIRA_ENABLE_WRITEStrue to register the write tools (comment, transition, assign, link); off by defaultLocal servers need a runtime on your machine (e.g. Node.js 18+). Config file locations differ by client; see your client’s docs.
Tools
10Saved tool list · updated 4d ago
| Parameter | Type | Required | Description |
|---|---|---|---|
| depth | enum | No | basic = the issue only; full = also parent, epic, linked issues, subtasks and commentsbasic | full |
| issueKey | string | Yes | Jira issue key (e.g. PROJ-123) |
| includeConfluence | boolean | No | Also search Confluence for related pages and fold them into the result (implies full depth) |
Details
From the project README
An MCP server and agent skill for Jira Cloud. Start a ticket with a ready-to-build brief,
check your branch against its acceptance criteria before the PR, and hand off — from your coding agent.
How it works
1 · Start a ticket
"I'm starting ABC-1, get me ready."
Readiness, blockers, acceptance criteria, next steps and a branch name — before you write a line.
2 · Verify before the PR
"Check my branch against ABC-1 before I open the PR."
Your local diff against the acceptance criteria and Definition of Done, with one verdict — ready_for_pr, needs_work or insufficient_evidence — and the files behind each criterion.
3 · Hand off
"Post the QA handoff to ABC-1."
QA handoffs, release notes, test plans and reviews. With write-back on, the agent shows you a preview and waits for your approval.
Install
You need Node.js 22+, your Jira Cloud URL, your Atlassian email and an API token.
Claude Code — plugin with the server and the skill
/plugin marketplace add teransarathchandra/jira-mcp-server
/plugin install jira-delivery@jira-delivery
It asks for your site, email and token (the token goes to secure storage). Without the plugin:
claude mcp add jira-delivery \
-e JIRA_BASE_URL=https://your-domain.atlassian.net \
-e [email protected] \
-e JIRA_API_TOKEN=your-api-token \
-- npx -y jira-delivery-mcp
Cursor
Click Add to Cursor and replace the three placeholder values, or add the JSON below to ~/.cursor/mcp.json.
VS Code (GitHub Copilot agent mode)
Click Install in VS Code (or Insiders). VS Code asks for your site, email and token and stores the token as a secret.
Manually: add to .vscode/mcp.json — VS Code uses servers, not mcpServers:
{
"servers": {
"jira-delivery": {
"type": "stdio",
"command": "npx",
"args": ["-y", "jira-delivery-mcp"],
"env": { "JIRA_BASE_URL": "https://your-domain.atlassian.net", "JIRA_EMAIL": "[email protected]", "JIRA_API_TOKEN": "your-api-token" }
}
}
}
Windsurf ·
Claude Desktop · any MCP client
Add to the client's MCP config — Windsurf: Settings → MCP Servers; Claude Desktop: claude_desktop_config.json:
{
"mcpServers": {
"jira-delivery": {
"command": "npx",
"args": ["-y", "jira-delivery-mcp"],
"env": { "JIRA_BASE_URL": "https://your-domain.atlassian.net", "JIRA_EMAIL": "[email protected]", "JIRA_API_TOKEN": "your-api-token" }
}
}
}
⌨️ Codex CLI
In ~/.codex/config.toml:
[mcp_servers.jira-delivery]
command = "npx"
args = ["-y", "jira-delivery-mcp"]
env = { JIRA_BASE_URL = "https://your-domain.atlassian.net", JIRA_EMAIL = "[email protected]", JIRA_API_TOKEN = "your-api-token" }
Check your setup: with the three variables exported, npx jira-delivery-mcp doctor verifies the login and prints the custom-field ids your site needs (acceptance criteria, story points, epic). Other agents can add the jira-delivery skill, also attached to each release.
Tools
| Step | Tools |
|---|---|
| Start | jira_search_issues · jira_get_issue · jira_prepare_work_brief |
| Verify | delivery_check_changes · delivery_traceability_matrix |
| Hand off | delivery_generate · jira_add_comment* · jira_update_issue* |
| Context | confluence_search · confluence_get_page · delivery_project_patterns · delivery_generate_agent_pack |
* Opt-in with JIRA_ENABLE_WRITES=true. Every tool returns typed structuredContent; see the tool reference.
Safe by default
- 🔒 Local and private — runs on your machine over stdio; talks only to your Atlassian site; no telemetry.
- ✋ Read-only unless you opt in — and then every write is previewed, waits for your approval, is sent once and audited.
- 🧱 Ticket text is untrusted — fenced as data, scanned for prompt injection, never turned into instructions.
- 🗝️ Credentials stay in your environment — never logged; redacted from all output.
Documentation
Usage · Tools · Configuration · Security · Troubleshooting · Migrating from 1.x · Changelog
Contributing
Issues and pull requests are welcome — see CONTRIBUTING.md. Security reports: SECURITY.md.
MIT © Teran Sarathchandra
Comments
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.