.NET MCP server (transport stdio) for Confluence Cloud (REST v2) and Jira Cloud (REST v3).

• Scoped API Token + Basic Auth (email:token)
• Automatic cloudId resolution via https://<site>/_edge/tenant_info
• Confluence REST v2 via Atlassian gateway
• Jira REST v3 via Atlassian gateway

• .NET SDK 10.0+
• An Atlassian Cloud site (e.g. https://example.atlassian.net)
• An API token with appropriate scopes for Confluence and/or Jira

Copy the example and fill in your values:

cp src/Atlassian.McpServer/appsettings.exemple.json src/Atlassian.McpServer/appsettings.json

Example:

{
  "Atlassian": {
    "Site": "https://example.atlassian.net",
    "Email": "you@company.com",
    "ApiTokenConfluence": "your_confluence_token_scoped",
    "ApiTokenJira": "your_jira_token_scoped"
  }
}

Notes:

• Never commit appsettings.json (use .gitignore).
• If you only use one product (Confluence/Jira), leave the other token empty.

dotnet restore src/Atlassian.McpServer/Atlassian.McpServer.csproj
dotnet build   src/Atlassian.McpServer/Atlassian.McpServer.csproj -c Release
dotnet run     --project src/Atlassian.McpServer/Atlassian.McpServer.csproj -c Release

The MCP process listens on stdin/stdout.

In Codex config.toml, add an MCP server pointing to the Release DLL:

[mcp_servers.atlassian]
command = "dotnet"
args = [
  "/path/to/repo/tools/atlassian-mcpserver/src/Atlassian.McpServer/bin/Release/net10.0/Atlassian.McpServer.dll"
]

Then restart VS Code (or the Codex extension) to reload the config.

• confluence_list_spaces(keys?, limit)
• confluence_get_space_homepage_id(space_key)
• confluence_get_space_info(space_key)
• confluence_get_folder(folder_id, include_children)
• confluence_get_page(page_id, include_children)
• confluence_root_folders(space_key)
• confluence_upsert_page(space_key, parent_id?, page_id?, title, body_storage_html, status?, version_message?, subtype?)
  • Recommended Jira reference: insert a Jira macro in body_storage_html:

    <ac:structured-macro ac:name="jira">
      <ac:parameter ac:name="key">KEY</ac:parameter>
    </ac:structured-macro>

• jira_list_projects(query?, start_at, max_results, order_by?)
• jira_get_project(project_id_or_key)
• jira_get_issue(issue_id_or_key, fields_csv?, expand_csv?)
• jira_search_jql(jql, max_results, next_page_token?)
• jira_create_issue(project_key, issue_type, summary, description?, labels_csv?, parent_issue?)
• jira_delete_issue(issue_id_or_key, delete_subtasks)

• confluence://spaces
• confluence://page/{pageId}
• confluence://folder/{folderId}
• confluence://root-folders/{spaceKey}
• jira://projects
• jira://project/{projectIdOrKey}
• jira://issue/{issueIdOrKey}
• jira://issue/createmeta/{projectKey}
• jira://issue/editmeta/{issueIdOrKey}

• Logs are sent to stderr to avoid polluting the JSON-RPC stream.
• The server uses ModelContextProtocol (C#). If needed:

  dotnet add package ModelContextProtocol --prerelease