Introduction
What Lific is, how its interfaces share one database, and which page to read first for your goal.
Lific is an issue tracker you run yourself, built to be used by coding agents as much as by people. One binary contains the server, a web UI, a command-line interface, a REST API and an MCP server. Everything it knows lives in one SQLite database and a directory of attachments on a machine you control. There is no hosted Lific service and no account to create anywhere else.
Issues, pages and plans have short readable identifiers such as APP-42, APP-DOC-3 and APP-PLAN-2. They are meant to be typed into a prompt, pasted into a commit message, or read aloud.
One database, four ways in
An instance is one database. The web UI, the CLI, the REST API and MCP all read and write that same data, so an issue an agent creates over MCP shows up in the browser immediately, and a status you change on the board is what the agent sees on its next call.
| Interface | Use it for | Needs the server running |
|---|---|---|
| Web UI | Triage, boards, reading and writing pages, account and project settings | Yes |
| MCP | Letting a coding agent plan, track and document its work | Not for local stdio, which opens the database directly. Yes for HTTP and for remote stdio. |
| CLI | Setup, administration, scripts, backups and recovery | No for local commands. Yes with --backend http. |
| REST API | Integrations and custom clients | Yes |
Separate instances never synchronize or merge. If you need a project somewhere else, move it with a project archive.
Where to start
Trying it on your own machine. Read Installation, then follow the Quickstart. You will have a running tracker and a connected agent in a few minutes.
Giving a coding agent a tracker. Connect agents writes the MCP configuration for your AI tools. Repository bindings then let an agent working in a checkout find its project without being told.
Running it for a team. Read Self-hosting for deployment, accounts and backups, Configuration for every setting, and Upgrading before you install a new version.
Building on it. The MCP tools, REST API and CLI references list every operation with its parameters. Git hooks closes issues from commit messages.
Learning the model. Concepts explains projects, issues, pages, plans and how they relate.
If you are an agent reading this
These docs are also published as plain text. /llms.txt is an index of every page, and /llms-full.txt is the whole documentation in one file.
A few conventions hold across the MCP tools and are worth knowing before your first call:
- Refer to things by identifier:
APP-42for an issue,APP-DOC-3for a project page,DOC-3for a workspace page,APP-PLAN-2for a plan. Comments use their numeric id. - A failed tool call returns text that starts with
Error:and says what to change. It is not a protocol error, so read the text. - For a small change to a long description or comment, use
edit_issue,edit_pageoredit_commentwithold_stringandnew_string. They replace one exact string and leave the rest alone, so you do not have to resend the whole body. - List results are paged. When more exist, the output ends with the
offsetto pass on the next call. - Call
list_issueswithworkable=trueto get open issues whose blockers are all done.
Working with Lific as an agent expands on these conventions. The MCP overview covers transports and authentication, the tool reference documents every parameter, and Limits lists every size and rate limit you can run into.