mcp-jira-confluence-corp

A Model Context Protocol server for Atlassian Jira and Confluence, designed for corporate networks. It transparently bypasses system HTTP proxies for internal domains, handles self-signed certificates, and supports Personal Access Token (PAT) authentication.

Why this package exists

Generic MCP servers like mcp-atlassian work well against Atlassian Cloud but struggle inside many corporate environments where:

1. A global HTTP_PROXY or HTTPS_PROXY (typically an SSH tunnel for a cloud API) silently intercepts on-prem traffic and causes timeouts.

2. Internal Jira and Confluence instances use self-signed TLS certificates.

3. The only available credential is a Personal Access Token rather than a username/password pair.

mcp-jira-confluence-corp addresses all three by:

• Clearing HTTP_PROXY-style environment variables on client construction.

• Listing internal domains that should bypass any proxy (via the MCP_BYPASS_DOMAINS environment variable or ProxyConfig.bypass_domains).

• Disabling SSL verification by default (configurable via verify_ssl).

• Accepting PATs through JIRA_PERSONAL_TOKEN / CONFLUENCE_PERSONAL_TOKEN.

Related MCP server: Confluence MCP Server

Installation

pip install mcp-jira-confluence-corp

Or run it without installing using uvx:

uvx mcp-jira-confluence-corp

Configuration

The server is configured through environment variables:

MCP client configuration

{
  "mcpServers": {
    "jira-confluence": {
      "command": "uvx",
      "args": ["mcp-jira-confluence-corp"],
      "env": {
        "JIRA_URL": "https://jira.example.com",
        "JIRA_PERSONAL_TOKEN": "<your-token>",
        "CONFLUENCE_URL": "https://confluence.example.com",
        "CONFLUENCE_PERSONAL_TOKEN": "<your-token>",
        "MCP_BYPASS_DOMAINS": "*.example.com,internal.corp",
        "HTTP_PROXY": "",
        "HTTPS_PROXY": ""
      }
    }
  }
}

Available tools

Jira (read-only)

Jira (write)

Confluence (read-only)

Confluence (write)

All tools support both markdown (default) and json response formats for human-readability vs. programmatic consumption.

Programmatic usage

JiraConfluenceClient is a thin wrapper around atlassian-python-api and is fully usable outside of the MCP server.

from mcp_jira_confluence import JiraConfluenceClient

client = JiraConfluenceClient()

issue = client.get_issue("PROJ-123")
print(issue["fields"]["summary"])

page = client.get_page("123456")
print(page["title"])

client.update_page(
    page_id="123456",
    title="Release notes",
    body="<p>Updated content in Confluence Storage Format.</p>",
)

field_map = client.build_field_map(project_key="PROJ", issue_key="PROJ-123")
for row in field_map["rows"]:
    print(row["field_id"], row["field_name"], row["schema_type"])

Customizing proxy bypass

from mcp_jira_confluence import JiraConfluenceClient, ProxyConfig

proxy_config = ProxyConfig(
    bypass_domains=["*.example.com", "internal.corp"],
    verify_ssl=False,
)
client = JiraConfluenceClient(proxy_config=proxy_config)

Development

git clone https://github.com/mshegolev/mcp-jira-confluence-corp.git
cd mcp-jira-confluence-corp
pip install -e ".[dev]"

pytest tests/
ruff check src/
black --check src/

Releasing

This project uses PyPI Trusted Publishing, so no API tokens are stored in GitHub.

1. Bump the version in pyproject.toml.

2. Update CHANGELOG.md.

3. Tag the release: git tag v0.1.1 && git push --tags.

4. Create a GitHub release — the publish workflow runs automatically.

License

MIT

Author

Mikhail Shegolev — mshegolev@gmail.com