gadak — Claude Code MCP server: SQL over Jira and Confluence
Copies the Jira Cloud projects and Confluence spaces you pick into a single SQLite file on your disk. From there you query it with SQL, full-text search or a single issue view, offline.
Open on GitHubWhat gadak does
gadak pulls chosen Jira projects and wiki spaces, including comments, change history and pages, onto your machine. A single SQLite database holds them, indexed as one. The MCP server then answers agent questions from that file instead of paging through REST calls.
The gain shows up in aggregates. JQL has no GROUP BY, so counting open issues per epic through the API means fetching rows and tallying them yourself; the mirror returns the count in one query. You can reach the same data from a desktop app, a browser tab started with gadak serve, the CLI, or MCP. Writes land in Jira first.
You need an Atlassian API token created without scopes. It covers both Jira and Confluence on one site, and the mirror shows exactly what your account shows. Scope is set at setup time with gadak init --projects ENG,PROD --spaces ENG; leave out --spaces and the wiki stays off. Jira Server and Data Center use gadak init --server with a Personal Access Token, and the wiki is unavailable there.
What it helps with
- count open issues per epic in one query instead of paging the API
- search comments and wiki pages by free text
- read an issue's change history without calling Jira
- keep issues and docs in one local database for custom reports
Example prompts for the agent
- Using gadak, count unresolved issues per epic in the ENG project
- Search the local gadak mirror for comments that mention billing
- Show the history of ENG-412 from the gadak mirror
How to install gadak
Install
The MCP registry has no launch command for it: see the repository README
Step by step, with commands for each platform: How to add an MCP server to Claude Code, Codex, Cursor, VS Code, Gemini CLI and more
Questions about gadak
How do I install gadak in Claude Code?
The plugin marketplace does not carry it — install gadak itself and wire it up as an MCP server. On macOS with the desktop app and CLI: brew install --cask midagedev/tap/gadak. CLI only on macOS or Linux: brew install midagedev/tap/gadak-cli.
What does gadak do?
It keeps a local copy of selected Jira Cloud projects and Confluence spaces in SQLite and serves it to your agent over MCP, in a browser or from the CLI. Reads come from disk; writes go to Jira first.
What do I need before connecting?
An Atlassian API token created without scopes, plus gadak init naming the projects. For Jira Server and Data Center, use a Personal Access Token and the --server flag instead.
