jira-mcp

tsantana84/jira-mcp

3.3

If you are the rightful owner of jira-mcp and would like to certify it and/or have it hosted online, please leave a comment on the right or send an email to henry@mcphub.com.

Jira MCP Server is a specialized server for interacting with Jira Cloud, focusing on issues management through a Model Context Protocol (MCP) interface.

Tools
1
Resources
0
Prompts
0

jira + confluence mcp servers (cloud)

simple, stdio-based mcp servers for atlassian jira and confluence (cloud, api token auth). no webhooks. focus on issue search, boards, and confluence search, plus a minimal "reports" server for daily briefs.

example: ai-powered dependency analysis

turn complex jira tickets into actionable implementation plans in 3 simple steps:

step 1: analyze jira dependencies

in gemini cli, paste:

run dependency analysis on DMD-11937 with:
- depth: 3
- include confluence docs updated in last 12 months
- save to jira_analysis.json

what you get: jira ticket context, dependency graph, blocker analysis, confluence docs, and a ready-to-use prompt for code analysis

step 2: analyze related code

copy the suggested_prompt from jira_analysis.json (replace {{YOUR_GITHUB_ORG}} and {{YOUR_GITHUB_REPO}}), then paste it into claude or gemini in your repository.

note: tell the ai to wait if it hits github rate limits - accuracy over speed for this report.

what you get: related prs, commits, implementation patterns, cross-repo dependencies with confidence scores - saved to code_analysis.json

step 3: synthesize implementation plan

in claude or gemini, run the synthesis prompt:

use the SYNTHESIS_PROMPT.md template with @jira_analysis.json and @code_analysis.json

what you get:

  • tech lead context: executive summary, effort estimate, risk assessment
  • developer guide: step-by-step implementation plan with code examples, testing strategy, deployment plan
  • correlation analysis: confidence-scored matches between jira context and code findings

output: synthesis_analysis.json - ready to paste into ticket descriptions or hand to developers

quick setup

prerequisites

install

git clone https://github.com/your-org/jira-mcp.git
cd jira-mcp
npm install
npm run build

configure for gemini cli (recommended)

# step 1: set your credentials (interactive)
npm run setup:gemini

# step 2: configure gemini cli
npm run gemini:config

ready to use! your gemini cli is now connected to jira and confluence.

manual mcp client configuration

if not using gemini cli, add to your mcp client config (claude desktop, cline, etc.):

"mcpServers": {
  "jira-min": {
    "command": "node",
    "args": ["/absolute/path/to/jira-mcp/scripts/minimal-server.mjs"],
    "cwd": "/absolute/path/to/jira-mcp",
    "transport": "stdio",
    "env": {
      "JIRA_BASE_URL": "https://your-site.atlassian.net",
      "JIRA_EMAIL": "you@example.com",
      "JIRA_API_TOKEN": "your-api-token"
    }
  }
}

important: use absolute paths, not relative paths.

available tools

jira-min server

  • jira_list_issues - search via jql with pagination
  • jira_get_issue - fetch single issue with full details
  • jira_dependency_analysis - comprehensive dependency analysis (recommended!)
  • jira_issue_relationships - traverse dependency graph
  • jira_get_changelog - status transitions, reassignments
  • jira_list_projects - list accessible projects
  • jira_list_boards - list boards for projects
  • jira_board_issues - get issues from specific boards
  • jira_find_similar_tickets - discover context from historical tickets
  • confluence_get_page - get page with ancestors/breadcrumbs
  • jira_issue_confluence_links - extract confluence links from issue
  • confluence_page_jira_links - extract jira keys from page

confluence-min server

  • confluence_search_pages - search pages using cql

reports-min server (jira + confluence combined)

  • ops_daily_brief - daily summary (last 24h, new issues, transitions, blocked tickets)
  • ops_shift_delta - what changed since you logged off
  • ops_jira_review_radar - issues waiting for review (stale prs)

common usage patterns

search jira issues

jql: "project = ABC AND status != Done ORDER BY updated DESC"
fields: "summary,assignee,comment"
includeComments: true

analyze dependencies for blocked ticket

issueKey: "PROJ-123"
depth: 3
autoDiscover: true

search confluence for architecture docs

cql: "text ~ 'microservices' AND (title ~ 'architecture' OR title ~ 'design') AND space = ENG"
limit: 10

get daily brief for last 24 hours

from: "2025-10-08T12:00:00-03:00"
to: "2025-10-09T12:00:00-03:00"
projects: ["ABC", "DMD"]
labelsBlocked: ["Blocked", "Impeded"]

advanced workflows

dependency analysis (3-stage process)

stage 1: jira/confluence discovery analyzes ticket dependencies, extracts confluence docs, identifies blockers output: jira_analysis.json with suggested code search prompt

stage 2: code analysis uses suggested_prompt from stage 1 with github cli in your repository searches for related prs, commits, implementation patterns output: code_analysis.json with structured findings

stage 3: synthesis correlates both analyses with confidence tracking generates tech lead context + developer implementation guide output: synthesis_analysis.json

see: for detailed workflow see: for stage 3 template

using with gemini cli

gemini cli is google's terminal-based ai assistant. connect this jira mcp server to:

  • ask questions about jira projects in plain english
  • search issues, boards, confluence pages from your terminal
  • get daily reports through conversational queries
  • combine jira data with github in a single ai session

detailed setup: includes: advanced configuration, combining with github mcp, jql patterns, troubleshooting

testing connectivity

# verify jira connection
npm run ping

# or manually
JIRA_BASE_URL=https://your-site.atlassian.net \
JIRA_EMAIL=you@example.com \
JIRA_API_TOKEN=your-token \
npm run start:jira-min

architecture

where things live

  • minimal servers: scripts/minimal-server.mjs, scripts/confluence-minimal-server.mjs, scripts/report-minimal-server.mjs
  • jira client: src/jira/client.ts (http + retries, agile + core apis)
  • issue normalization: src/jira/issues.ts (adf to plain text, field mapping)
  • mcp integration: src/mcp/tools.ts (tool registration, input/output schemas)
  • schemas: src/schemas.ts (zod schemas for validation)

confidence tracking

all analysis outputs include explicit confidence scores:

  • high (0.8-1.0): direct evidence (ticket id in pr, exact component match)
  • medium (0.5-0.8): inferred match (similar patterns, related terms)
  • low (0.0-0.5): weak connection (keyword match only, old references)

gaps are flagged explicitly rather than filled with assumptions.

troubleshooting

common issues:

  • tool list caching in mcp clients
  • logs not visible (use stderr for debugging)
  • 404 on confluence (missing /wiki in base url)
  • jira 410 errors (search api migration)
  • mcp server name collisions

see: for solutions

security

  • keep your atlassian api token secret
  • rotate it if it appears in shell history or logs
  • use environment variables, not hardcoded tokens in configs

license

mit