M
MCP Directory CURATED DIRECTORY
911 MCP servers indexed
中文
← Back to directory
M

MCP Atlassian Server

Community
MCP server connecting AI assistants to Jira and Confluence
How to install ↓ Compare GitHub source repository ↗
Category
Collaboration #27 of 55
Stars
★ 52 Popular
Runtime
Node.js
Credentials
API key / credential required
License
MIT
Last commit
⚠
54FMRS · D

A feature-rich third-party MCP server (48 features) covering major Jira and Confluence read/write operations, notably optimized for Cline with thorough install docs; not officially maintained, so token permissions must be managed carefully.

Strongest · Maintenance 14/20 Weakest · Reliability 8/20

Reliability
8/20
Security and permissions
8/20
Maintenance
14/20
Documentation
13/20
Setup experience
11/20
Why each score
Reliability 8/20
The README describes 48 features across Jira/Confluence resources and tools and includes a request-flow diagram, but no manifest was cached and there is no description of CI workflows or committed tests to verify the MCP init/tool-list handshake or that declared behavior matches real behavior. Per the static calibration rule, reliability is capped below 12 absent verifiable execution evidence (real CI plus tests), so a low-to-mid score is warranted.
Security and permissions 8/20
The README includes a dedicated 'Security Note' section warning that the API token inherits full user permissions, advising against sharing tokens, and cautioning about exposing config files to LLMs — a positive signal. However, the feature set includes numerous destructive operations (delete pages, delete comments, delete filters, remove dashboard gadgets) with no described confirmation flow, least-privilege scoping, or data-boundary controls for these actions, and no stated limits on external network calls. The absence of these controls for destructive operations warrants a deduction.
Maintenance 14/20
The project maintains a detailed CHANGELOG showing version progression (2.0.1 through 2.1.1 with expanding features), uses the MIT license, is not archived, and shows a healthy issue count (4 open) relative to 53 stars, indicating ongoing activity. However, the README does not describe a clear security-response channel or formal release/tagging cadence, so governance signals are incomplete, preventing a top score.
Documentation 13/20
Documentation is reasonably layered: README, CHANGELOG, a dedicated install guide (llms-install.md), and a technical reference (docs/introduction/resources-and-tools.md), supported by architecture and sequence diagrams. However, only the README body is directly reviewable here — parameter-level tool documentation, rate-limit/cost disclosures, and troubleshooting content are only referenced via links, not directly verifiable, so points are deducted for unreviewed evidence.
Setup experience 11/20
The README offers a one-line Smithery install command and points to a detailed llms-install.md guide covering prerequisites, Node.js setup, and API token acquisition, suggesting a path from environment setup to a working connection. Per the static calibration rule, without CI or test evidence confirming reproducibility, and since the full installation steps are not directly visible in this review (only referenced), setup cannot be scored higher.

Static review · not runListed 2026-09-07

Read the FMRS scoring method →

Fit and risk

What it can accessUses the networkChanges third-party account data

Best for

  • Developers who use Jira and Confluence daily with Cline
  • Individuals wanting to integrate Atlassian into AI workflows without context switching
  • MCP-compatible client users seeking quick Atlassian integration

Not for

  • Teams requiring enterprise-grade official support or Atlassian-maintained assurance
  • Users without an MCP-compatible client
  • High-compliance environments needing strict centralized token governance

Required permissions

  • Requires an Atlassian API token that inherits all permissions of its creator
  • Tools can perform create, update, and delete mutations on Jira and Confluence data
  • Runs locally and needs access to Atlassian cloud APIs

Risks and side effects

  • The API token inherits all of the creator's permissions, so a leak is high-impact
  • Never ask LLMs to analyze config files containing your token
  • Third-party, non-official project; Atlassian API changes may require maintenance
  • AI-driven mutations (deleting pages, transitioning issues) risk unintended changes

Setup

Before you start

Runtime：Node.js

Follow the llms-install.md guide in the repository (formatted for AI assistants to parse); you need a Node.js environment and an Atlassian API token. Alternatively install via Smithery: npx -y @smithery/cli install @phuc-nt/mcp-atlassian-server --client claude. Cline users can simply ask Cline to install this server.

Check that it works

Check the client's registered resources and tools (e.g. in Cline) for jira://issues resources and tools like createIssue; then ask "Find all my assigned issues" — a populated answer confirms the connection.

Troubleshooting

  1. Check that the Atlassian API token is correctly configured and not expired
  2. Confirm your Node.js environment meets system requirements
  3. Verify the Cline client configuration step by step against llms-install.md
  4. Confirm your account has access to the target Jira site and Confluence spaces

Things to try

Once connected, you can ask your AI assistant things like:

  • Find all my assigned Jira issues
  • Create a new Jira issue about a login bug
  • List all pages in my Confluence space
  • Add a comment to a specific Confluence page

Use cases

Have an AI assistant query and filter issues assigned to you in Jira
Create, update, and transition Jira issues via natural language
Create, update, and delete Confluence pages
View and manage Jira boards, sprints, and filters
Manage Confluence page comments and attachments

Supported clients

Cline
Claude DesktopPartial support
CursorPartial support

Listed from the project's documentation, not tested by this site.

Overview

MCP Atlassian Server (by phuc-nt) is a Model Context Protocol server that connects AI agents such as Cline, Claude Desktop, or Cursor to Atlassian Jira and Confluence, letting them query data and perform actions through a standardized interface. Version 2.0.1 expanded features from 21 to 48, using Jira API v3 and Confluence API v2, with full Agile/Scrum workflow support, page version management, attachments, and comments. The server distinguishes read-only Resources from action Tools. It is primarily optimized for Cline, designed local-first, and released under the MIT license.

Similar servers

Compare with other MCP servers

Source revision 6ca56e82e7f8 Data synced 2026-10-19 Read the FMRS scoring method