What it does

Lix is an in-process version control library you import into your app or agent worker. It gives you branches, transactions, rollback, and immutable history for non-code files—PDFs, spreadsheets, JSON, CSV—backed by whatever storage you already use (SQLite, Postgres, S3, Cloudflare).

The interesting bit

Git sees an XLSX file as a binary blob and shrugs. Lix parses structure: it can tell you that cell B3 in the “orders” sheet flipped from “pending” to “shipped,” and your agent can query that change with SQL instead of re-reading the whole file. The “semantic change” layer is the real bet here.

Key highlights

• Runs in-process; no daemon, no protocol to speak
• ACID transactions across files, blobs, and history in one go
• Parallel branches without Git-style worktree juggling
• SQL interface over lix_change table—filter by branch, schema, entity, time
• Pluggable backends: start in-memory, graduate to Postgres or S3

Caveats

• CLI and file plugins for DOCX/XLSX/PDF are still on the roadmap (v0.7–v0.8); the SDK is there, but format-specific smarts are pending
• Python, Rust, and Go bindings are tracked as open issues, not shipped yet
• Merge conflicts and working changes/checkpointing are not implemented until v0.9–v0.10

Verdict

Worth a look if you’re building agent tools or collaborative apps that version non-code assets and want programmatic query access to history. Skip it if you need a mature CLI or deep format support today—you’re buying into a roadmap.