Understand the code you ship.

Tate is Anki for your codebase. Spaced repetition that closes the gap between "this code exists" and "I can defend this."

Install Documentation

$ tate review

src/auth/login.ts::authenticate

Review #3 · Last reviewed 8 days ago

What happens when the token is expired but the

refresh token is still valid?

[1] Again [2] Hard [3] Good [4] Easy

> 3

Next review: April 5

0 cards remaining today.

Tests verify behavior. Code review checks style. Git blame tracks authorship. Nothing verifies that anyone actually understands the code that shipped.

That gap is invisible until it isn't. A teammate leaves and nobody can touch their module. AI writes a function and you ship it because the tests pass. You copy a pattern from Stack Overflow and move on. The code works, but you can't defend it.

#Code you can't defend

The AI wrote it. The tests pass. You shipped it. Could you rewrite it from scratch?

Someone left. Their module is yours now. You've read it twice and still can't explain it.

You vendored a library. Your app depends on three functions you've never opened.

You wrote it six months ago. It works. You have no idea why.

#Built for developers

SM-2 Algorithm

The same proven scheduling behind Anki. You review what's fading, skip what you know. ~15 min/day.

26+ Languages

Track individual functions, not just files. Tree-sitter extracts symbols from Rust, TypeScript, Python, Go, and more.

Git Hook Auto-Add

AI-assisted commits automatically populate your deck. No bookkeeping required.

Agent Skill

Your AI writes code, then writes the review card. Drop-in skill for Claude Code, Cursor, and others.

Change Detection

Code changed? Card resets. Understanding is never permanent. It's earned, maintained, and re-earned.

Plain Text Deck

One entry per line in .tate/deck. Human-readable, grep-friendly, no lock-in.

#How it works

Code enters your deck

Manually, via git hook, or automatically by your AI coding agent.

Questions test understanding

Not "what does this do?" but "what breaks if this assumption stops being true?"

Review on schedule

Grade your understanding: Again, Hard, Good, or Easy. One keypress.

Spaced repetition schedules you

SM-2, the algorithm behind Anki. Reviews stay bounded at ~15 min/day.

Code changes, understanding resets

Modified code is detected automatically. You re-earn understanding.

Everyone is building tools to write more code faster. Nobody is building tools to make sure you still understand what ships.

Understanding decays. Tate brings it back.

$ cargo install tate-cli

$ cd your-project && tate init

$ tate review