Bridge-MCP

A unified Model Context Protocol (MCP) server connecting AI assistants to Jira, GitLab, and Confluence — issues, merge requests, pipelines, and docs from one server.

• Jira — issues, JQL search, comments, attachments, inline media

• GitLab — projects, MRs, issues, pipelines, repo, releases (via your instance's native GitLab MCP endpoint, GitLab 18.x+)

• Confluence — pages, spaces, search, child pages, attachments, comments

Works with self-hosted / private / VPN instances; per-service SSL toggle and Personal Access Token auth.

Use it (hosted)

The server runs in the cluster and is reached by URL (how to deploy).

• URL: https://bridge-mcp-baly-internal.apps.private.okd4.teh-1.snappcloud.io/mcp

• Auth: header Authorization: Bearer <token>

• Requires VPN (e.g. FortiClient) for private instances.

All callers share the server-side identity — every request uses the deployer's Jira/GitLab/Confluence tokens.

Claude Code

claude mcp add --transport http bridge-mcp <url> --header "Authorization: Bearer <token>"

Claude Desktop (bridge the URL with mcp-remote)

{ "mcpServers": { "bridge-mcp": {
  "command": "npx",
  "args": ["-y", "mcp-remote", "<url>", "--header", "Authorization: Bearer <token>"]
}}}

VS Code / Cursor / Copilot

{ "servers": { "bridge-mcp": {
  "type": "http", "url": "<url>", "headers": { "Authorization": "Bearer <token>" }
}}}

Gemini CLI (~/.gemini/settings.json)

{ "mcpServers": { "bridge-mcp": {
  "httpUrl": "<url>", "headers": { "Authorization": "Bearer <token>" }
}}}

Codex CLI (~/.codex/config.toml)

[mcp_servers.bridge-mcp]
url = "<url>"
http_headers = { Authorization = "Bearer <token>" }

Confirm tools loaded: /mcp in-session, or your client's tools panel.

Related MCP server: Personal JIRA MCP

Usage

Ask in plain language — the assistant picks the tool:

• "Show Jira issue PROJ-1234 and its comments."

• "Search Jira for open bugs assigned to me in PAY."

• "Why did the latest pipeline on main fail? Show the failing job log."

• "Show the diff and notes for merge request !57."

• "Find Confluence pages about the deployment runbook."

Tools

• Jira (5): get_jira_issue, search_jira_issues, get_issue_comments, get_issue_attachments, get_issue_media

• GitLab: dynamic — provided by your instance's MCP endpoint (projects, MRs, issues, pipelines/jobs, repo, releases/tags, members, labels). Run /mcp to see your set.

• Confluence (7): get_confluence_page, search_confluence_pages, get_space, list_spaces, get_page_children, get_page_attachments, get_page_comments

Full parameter docs: TOOLS.md.

Deployment

Container serving streamable-HTTP (src/http_server.py, port 8080, endpoint /mcp, health /healthz /readyz). Deployed to OKD namespace baly-internal via Helm + GitLab CI.

1. Setup — push repo to the internal GitLab (with OKD runners); set masked CI/CD vars: BUILDER_TOKEN, BRIDGE_MCP_AUTH_TOKEN, JIRA_PAT, GITLAB_PAT, CONFLUENCE_TOKEN. Service URLs live in helm/bridge-mcp/values-staging.yaml.

2. Deploy — run pipeline jobs in order: push:staging → helm_publish:staging → deploy:staging. Tokens injected via --set-string (no secret committed).

3. Result — the OpenShift Route publishes the URL above.

Container env vars: BRIDGE_MCP_AUTH_TOKEN (gates /mcp), JIRA_*, GITLAB_*, CONFLUENCE_*, PORT. Empty token = that service disabled.

Local development

python -m venv .venv && source .venv/bin/activate
pip install -r requirements.txt
cp .env.example .env          # fill in the services you use
python test_connectivity.py   # check reachability

Tokens: Jira → Settings → Personal Access Tokens · GitLab → Access Tokens (api, read_api, read_repository) · Confluence → Atlassian API Tokens. Self-signed certs: set *_VERIFY_SSL=false.

Run over stdio (local MCP clients launch this automatically):

{ "mcpServers": { "bridge-mcp": {
  "command": "/path/to/bridge-mcp/.venv/bin/python",
  "args": ["/path/to/bridge-mcp/src/bridge_mcp_server.py"]
}}}

Claude Code shortcut: claude mcp add bridge-mcp -- /path/to/.venv/bin/python /path/to/src/bridge_mcp_server.py.

Run over HTTP locally: BRIDGE_MCP_AUTH_TOKEN=secret PORT=8080 PYTHONPATH=src python src/http_server.py.

Troubleshooting

• Service disabled / no tools — credentials set? GitLab needs <GITLAB_URL>/api/v4/mcp reachable + GitLab 18.x+.

• 401 / connection failed — VPN connected, token valid, Authorization: Bearer header present (hosted).

• SSL error — set *_VERIFY_SSL=false.

License

MIT