Confluence MCP Server: Connect Claude Code, Cursor, and VS Code in 2 Minutes (2026)

You can connect Confluence to your AI coding agent in a couple of minutes by pointing it at Atlassian's official remote MCP server and approving an OAuth prompt. The moment you do, though, you hit a problem Jira does not have as badly: Confluence is where docs go to get stale, and a Confluence MCP server will hand your agent the half-finished drafts, the duplicate pages, and the superseded runbooks with exactly the same confidence as the one page that is actually true.

Updated September 10, 2026 for the Rovo MCP v2 endpoint Atlassian made generally available on September 8, the March 2027 v1 cutover, current Rovo credit and licensing terms, and the community server's latest tool count.

That staleness is not a small problem. Atlassian's own research found Fortune 500 teams waste roughly 2.4 billion hours a year searching for information, with over a quarter of the workweek lost to the hunt (Atlassian State of Teams 2025, 2025), and duplicated work follows from the fragmentation. An access protocol pointed at that mess inherits all of it. What helps underneath is a context engine that reconciles conflicts across your docs, so the page your agent gets back is the one that is actually current.

This guide gives you the current config for each client, updated for the v2 endpoint, plus the official-versus-self-hosted decision. It also covers the two things every Confluence connector walkthrough skips: the token tax you pay on every turn, and the gap between reading a page and knowing whether that page is current.

What is a Confluence MCP server, and what can your agent do with it?#

The Model Context Protocol now runs more than 10,000 active public servers and draws over 97 million monthly SDK downloads (Model Context Protocol, 2025). MCP is the open standard that lets an AI client call external tools, and a Confluence connector exposes Confluence operations as those tools.

In practice, your agent can search pages, read a page's full body, list spaces, post a comment, or create and update content, all through tool calls instead of copy-pasting. It acts under your own Confluence permissions, so it cannot read a space you cannot.

Why does that beat a plain API script? The agent decides when to call which tool, mid-task, without you wiring the sequence. Ask it to draft a release note, and it searches the space and reads the linked design doc on its own.

Two servers matter: Atlassian's official remote one and the community self-hosted one. Which you pick depends on whether you run Cloud or on-prem.

Which Atlassian MCP server should you use, official or community?#

There are two real options, and the choice is close to binary. Atlassian's official Rovo MCP Server reached general availability on February 4, 2026: remote, hosted, Cloud-only, and built on OAuth 2.1 (Atlassian, 2026). The community route is self-hosted and reaches on-prem.

The official server respects each user's permissions and, since the v2 release went generally available on September 8, 2026, spans Jira, Confluence, Jira Service Management, Bitbucket Cloud, Compass, Loom, Goals, and Projects (Atlassian developer changelog, 2026). If your team lives in Atlassian Cloud, this is your answer. (It is the same server family covered in our Jira MCP setup guide; the Confluence tooling rides along with it.)

You do not need a separate Rovo license. Rovo ships with Standard, Premium, and Enterprise Cloud plans, and connecting to the MCP server is free; only enriched Teamwork Graph calls draw down Rovo credits, most at 1 to 10 credits per call, with overage billing starting December 3, 2026 (Atlassian licensing, 2026; Rovo usage limits, 2026).

The community sooperset/mcp-atlassian project carries roughly 5,900 GitHub stars and ships 98 tools across Jira and Confluence (GitHub, 2026). It runs in Docker and, unlike the official server, supports both Cloud and Confluence Server or Data Center v6.0+.

The rule of thumb: Cloud teams use the official server, Server or Data Center teams use the community one. The full side-by-side sits a couple of sections down.

How do you connect Confluence to Claude Code, Cursor, and VS Code?#

Two dates break the old guides. The legacy https://mcp.atlassian.com/v1/sse endpoint stopped being supported after June 30, 2026 (Atlassian Community, 2026), and on September 8, 2026 Atlassian made the v2 server generally available at https://mcp.atlassian.com/v2/mcp (Atlassian developer changelog, 2026). The /v1/mcp/authv2 endpoint still answers, but anything on v1 is switched to v2 tools automatically on March 1, 2027, so every snippet below uses v2.

Setup takes about two minutes once you have an Atlassian Cloud site and a supported client. If you copied config from an older walkthrough, the endpoint line is the thing to change.

VS Code#

Add the server to .vscode/mcp.json, or search @mcp Atlassian in the Extensions view and install it from the gallery:

json{"servers":{"atlassian-mcp-server":{"url":"https://mcp.atlassian.com/v2/mcp","type":"http"}}}

Cursor#

Add it to ~/.cursor/mcp.json (the Atlassian plugin on the Cursor Marketplace does the same thing in one click):

json{"mcpServers":{"Atlassian-MCP-Server":{"url":"https://mcp.atlassian.com/v2/mcp"}}}

On older Cursor builds that cannot complete a remote OAuth flow directly, bridge through npx mcp-remote@latest <url> instead.

Claude Code#

Run one command, then authorize:

bashclaude mcp add --transport http atlassian https://mcp.atlassian.com/v2/mcp

Then run /mcp inside Claude Code, or claude mcp login atlassian from the shell, to trigger the browser OAuth flow (Claude Code docs, 2026).

The first connection opens a browser-based OAuth 2.1 consent screen. Once you approve, the agent acts under your own Confluence permissions, so its reach matches your existing access. For capacity planning, Atlassian's product page lists 500 calls per hour on Free and 1,000 per hour on Standard, with Premium and Enterprise scaling up to 10,000 per hour (Atlassian, 2026), and the enriched Teamwork Graph tools draw 1 to 10 Rovo credits per call on top of that.

This is also where the connector count starts to grow. The Slack thread and the GitHub commit that overruled the Confluence page each need their own server, each with its own OAuth flow and its own schema. Unblocked covers Confluence, Jira, Slack, GitHub, Notion, and more through one MCP connection, and the permissions you set in Confluence and Jira carry through to the person and their agent.

How do you self-host a Confluence connector for Server or Data Center?#

If you run Confluence on-prem, the official server cannot help, because it is Cloud-only (Atlassian, 2026). The community sooperset/mcp-atlassian server is the supported route, covering Confluence Cloud and Confluence Server or Data Center v6.0+.

The workflow is straightforward. Run the server in Docker, then authenticate: an API token on Cloud, a personal access token on Server or Data Center, or OAuth 2.0 if you prefer. Once it is running, point your client at the local server URL exactly as you would any other MCP server.

The tradeoff is ownership. You host it and patch it, and you own the configuration that decides how much of its surface your agent sees. In return you get on-prem coverage the official endpoint cannot offer. What you give up is Atlassian support, since the project is community-maintained and labels itself as not official.

For a regulated team that cannot send page data to a hosted endpoint, that ownership is worth the operational work: container updates, credential rotation, and monitoring all land on you.

Official Rovo MCP vs the community server: how they compare#

Most Cloud teams will reach for the official Rovo server, which now spans eight Atlassian products, while on-prem teams have one option: the community server with its 98 tools (GitHub, 2026). The table below lines them up on the dimensions that actually decide the call.

The read-out: official for Cloud, community for on-prem. The row people miss is the cost both share. Each one makes your agent pay for its tool-definition schema on every turn, and neither can tell the agent which of two conflicting pages is the current one. Both also stop at Atlassian's walls, so the Slack thread or GitHub commit that changed the decision needs yet another server. Those are the next two sections.

Why does a Confluence MCP server cost more tokens than you think?#

Your agent pays for a server's tool definitions on every turn, before it reads a word of your request. Anthropic's engineering team reported agents processing hundreds of thousands of tokens before reading a request, with one code-execution pattern cutting a workload from 150,000 to 2,000 tokens, a 98.7% drop (Anthropic, 2025).

That tax is invisible because it never throws an error. It shows up as a smaller working budget. Community measurements of the official GitHub server put its tool definitions alone at roughly 42,000 to 55,000 tokens before the first prompt, around a fifth of a 200,000-token window. A broad Atlassian toolset, or the community server's 98 tools, is the same shape of cost. Atlassian has conceded the point: the v2 server hides dozens of tools behind discover and execute calls and lazy-loads them, which its own preview notes say cut up-front context window consumption by more than half (Atlassian Community, 2026).

The practical move is to run fewer, leaner servers and enable only the toolsets you actually call, so more of your context window is left for the work. Anthropic's guidance on context engineering makes the same case: bloated tool sets create ambiguous decision points, and a model's recall degrades as its context fills (Anthropic, 2025). For the full accounting of how that per-turn cost compounds across a stack of servers, the numbers add up faster than most teams expect.

Unblocked removes the multiplier. One MCP connection covers Confluence, Jira, Slack, GitHub, Notion, and more, so the agent carries a single schema for all of them. The context that comes back is scored and compressed server-side before it reaches the agent, so the tokens go to the answer, and the per-turn cost of a five-server stack drops to the cost of one.

Does a Confluence MCP give your agent understanding, or just the page?#

A connector returns the page text; it cannot tell the agent whether that page is true. Developer distrust of AI accuracy stood at 46% in the most recent Stack Overflow survey, against 33% who trust it, and 66% say they lose time fixing answers that are almost right (Stack Overflow, 2025). The 2026 edition closed for responses in June and has not published results yet.

A Confluence connector hands the agent a page's title, body, and last-edited date. It does not tell the agent which of three overlapping pages is authoritative, that the current architecture doc was quietly overruled in a Slack thread, or that half the space is abandoned drafts. Confluence is where staleness accumulates, and Atlassian's own data puts the time teams lose to searching at over a quarter of the workweek (Atlassian State of Teams 2025, 2025). An access protocol can pull a page back. Reconciling two pages that disagree is beyond it.

Dumping all three pages into the window does not help either. The NoLiMa benchmark found that when the answer shares little wording with the question, 11 of 13 long-context models fell below half their short-context accuracy by 32,000 tokens (Modarressi et al., NoLiMa, 2025). An agent handed three overlapping runbooks is in exactly that position: more text, weaker recall, and no signal about which page wins.

The reason a page exists, and the decision that superseded it, lives in the threads and pull requests around it. This is the gap between reading a page and understanding it. Closing it takes a context engine that resolves conflicts across your docs, unifying pages, threads, and pull requests into one synthesized answer: decision-grade context rather than raw page text. That is what Unblocked does: when a Confluence page disagrees with a newer Slack thread or a later commit, the contradiction is resolved automatically by recency, authority, and proximity, and every answer ships with source links so the agent, and the person behind it, can see which page won and why.

Unblocked combines all of the tools we use into one product, one single MCP. And the search is efficient: the context it gathers is relevant and recent. I haven't found another tool that does that as effectively as Unblocked does.

Jorden van Breemen — CTO, Advidi

That is the shift from access to understanding. Unblocked is the context engine for agentic software development: your agents already know the code, and a context engine beneath MCP gives them the rest, including which Confluence page to trust.

Frequently asked questions about Confluence MCP#

Here are the questions teams ask most about wiring Confluence into an agent, answered against the current product facts: the official server reached general availability on February 4, 2026, and is Cloud-only (Atlassian, 2026).

Is there an official server for connecting Confluence to an agent?#

Yes. Atlassian's Rovo MCP Server reached general availability on February 4, 2026, and its v2 endpoint followed on September 8, 2026 (Atlassian, 2026; Atlassian developer changelog, 2026). It is Cloud-only, uses OAuth 2.1, and covers Confluence alongside Jira, Jira Service Management, Bitbucket Cloud, Compass, Loom, Goals, and Projects. For Server or Data Center, the community sooperset/mcp-atlassian project is the route instead.

How do I connect Confluence MCP to Claude Code?#

Run claude mcp add --transport http atlassian https://mcp.atlassian.com/v2/mcp, then run /mcp inside Claude Code to authorize through the browser OAuth 2.1 prompt. The agent then searches and reads Confluence under your own permissions. Use the v2 endpoint: the legacy /v1/sse endpoint stopped working after June 30, 2026, and v1 switches to v2 tools on March 1, 2027.

How do I set up the Atlassian MCP server in Cursor?#

Add the v2 URL to ~/.cursor/mcp.json as {"mcpServers":{"Atlassian-MCP-Server":{"url":"https://mcp.atlassian.com/v2/mcp"}}} and approve the OAuth consent screen on first connection, or install the Atlassian plugin from the Cursor Marketplace. Older Cursor builds that cannot complete a remote OAuth flow should bridge through npx mcp-remote@latest pointed at the same URL.

Does the Atlassian MCP server work with Confluence Server or Data Center?#

The official server does not; it is Cloud-only (Atlassian, 2026). For Confluence Server or Data Center v6.0+, self-host the community sooperset/mcp-atlassian server in Docker and authenticate with a personal access token. It also supports Confluence Cloud if you want a single self-managed server across both.

Why doesn't my agent trust the Confluence page it read?#

Because a connector returns page text and nothing about whether the page is still true. It hands over a body and an edit date, but it cannot tell which of several overlapping pages is authoritative or whether a Slack decision quietly overruled it. Developer distrust of AI accuracy stood at 46% in the latest Stack Overflow survey (Stack Overflow, 2025) for related reasons. Unblocked closes that gap by shipping every answer with source links, so the agent can show which page, thread, or commit it relied on, and the permissions set in Confluence and Jira carry through to the person and their agent.

Is a Confluence connector enough, or do you need a context engine?#

A Confluence connector is enough to read pages under your own permissions. It is not enough when pages are stale or contradict each other, because it returns text without resolving which one is authoritative. A context engine like Unblocked unifies Confluence with Slack, Jira, GitHub, and Notion through one MCP connection and returns the reconciled answer, so the agent acts on the decision that is actually current.

Where the Real Context Lives#

If you are on Atlassian Cloud, connect the official remote server today; the setup runs about two minutes against the v2 endpoint. If you are on Server or Data Center, self-host sooperset/mcp-atlassian instead. If you want the wider field first, our roundup of the best Confluence and Jira MCP servers compares four options side by side. Either way, prune the toolset so your agent is not paying for 98 tools it never calls, a cost that can run to tens of thousands of tokens before the first prompt (Anthropic, 2025).

Then know the ceiling. A connector gets your agent the page. Whether that page is current, and what decision overruled it, lives in the threads and pull requests around it, where Atlassian's research says over a quarter of the workweek already disappears into searching (Atlassian State of Teams 2025, 2025). Unifying and reconciling those sources is what turns page access into page understanding, and it is the job Unblocked was built for: one MCP connection across Confluence, Jira, Slack, GitHub, and Notion, contradictions settled by recency, authority, and proximity, source links on every answer, Confluence and Jira permissions carried through to each person and their agent, and context compressed before it reaches the model. Keep the Atlassian connector for writes, and put Unblocked in front of it for the reads that decide what gets built. That is the difference between an agent that fetches a page and one that knows which page to believe.