ait¶
The Git workflow layer for AI coding agents — worktree isolation, attempt provenance, cross-session memory, reviewable promotion.
ait wraps the agent CLIs you already use — Claude Code, Codex, Aider,
Gemini CLI, Cursor — and turns each run into a reviewable attempt.
The agent edits an isolated Git worktree, ait records what happened,
and your main checkout stays untouched until you promote the result.
The package is named ait-vcs on PyPI and npm. The installed command is ait.
Problems ait solves¶
| Problem with agent coding today | What ait adds |
|---|---|
| A bad prompt rewrites half your repo before you notice | Each run lands in an isolated Git worktree — your root checkout never moves |
| The diff has no useful provenance — which prompt produced it? | Attempts link intent, command output, files, and commits in one record |
| Failed or partial runs leave your working copy half-broken | Bad attempts stay isolated; ait attempt discard removes them in one command |
| The next agent repeats investigation you already paid tokens for | Repo-local memory feeds prior attempts and commits to the next run |
| Two agents on the same task stomp each other | Each attempt has its own worktree — run N agents in parallel |
| Did the agent really fix it, or just claim it did? | Explicit ait attempt promote keeps speculative changes out of main until you decide |
| Cross-agent hand-offs lose every previous decision | Memory layer auto-imports CLAUDE.md, AGENTS.md, and prior attempts |
| Provenance tooling wants to ship your code to a SaaS | Metadata stays in .ait/ next to .git/ — harness daemon is local-only (Unix socket, no network), no telemetry |
| "Where's that prompt I wrote last month?" → grep shell history | Query attempts, intents, and commits with a structured DSL |
See the full deep-dive on each problem in Why ait.
ait is not another agent. It is the Git layer around the agents you
already trust.
Supported agents¶
Status¶
ait is alpha quality. It is intended for local dogfooding and early users
who are comfortable with Git workflows. Metadata is local to one repository
under .ait/; it is not synchronized across machines.