Description
Searching an issue in Jira means a click and a wait, a page turn and another wait; and a question like "how many issues are still open under each epic" is something JQL cannot express, so you count by hand. gadak takes another route: it mirrors the Jira projects and Confluence spaces you choose into one SQLite file on your machine, with issue titles, bodies, comments and wiki pages all in one search index, and searching it opens no connection.
Once it is a local file, the questions JQL cannot answer take one SQL query — a single GROUP BY runs in 22 ms over three thousand issues. Search is instant and works offline, reads never touch the network, and writes still go to Jira first with the local file just a cache you can delete and rebuild. The desktop app, a browser tab, the CLI and an SQL/MCP interface for AI agents all read that same file. Homebrew on macOS, the Microsoft Store on Windows, and a CLI on Linux that opens the same UI in a browser tab — one binary.
Written in Go, Apache-2.0.
Mirror into one SQLite: the selected Jira projects and Confluence spaces sync wholesale into a local SQLite file, with issue bodies, comments and wiki pages in one search index.
SQL for what JQL cannot do: run SQL over the local file — one GROUP BY answers over three thousand issues in 22 ms, and a join across change history is one query.
Instant offline search: search hits the local index with no network request and works offline, while writes go to Jira first and the file stays a rebuildable cache.
SQL/MCP for agents: the desktop app, browser, CLI and coding assistants read the same file over SQL/MCP, letting an agent query your backlog directly.
One binary, cross-platform: Homebrew on macOS, the Microsoft Store on Windows, and a Linux CLI that opens the same UI in a browser tab.
Once it is a local file, the questions JQL cannot answer take one SQL query — a single GROUP BY runs in 22 ms over three thousand issues. Search is instant and works offline, reads never touch the network, and writes still go to Jira first with the local file just a cache you can delete and rebuild. The desktop app, a browser tab, the CLI and an SQL/MCP interface for AI agents all read that same file. Homebrew on macOS, the Microsoft Store on Windows, and a CLI on Linux that opens the same UI in a browser tab — one binary.
Written in Go, Apache-2.0.
Features
Mirror into one SQLite: the selected Jira projects and Confluence spaces sync wholesale into a local SQLite file, with issue bodies, comments and wiki pages in one search index.
SQL for what JQL cannot do: run SQL over the local file — one GROUP BY answers over three thousand issues in 22 ms, and a join across change history is one query.
Instant offline search: search hits the local index with no network request and works offline, while writes go to Jira first and the file stays a rebuildable cache.
SQL/MCP for agents: the desktop app, browser, CLI and coding assistants read the same file over SQL/MCP, letting an agent query your backlog directly.
One binary, cross-platform: Homebrew on macOS, the Microsoft Store on Windows, and a Linux CLI that opens the same UI in a browser tab.
