prepcli
stop re-explaining.
start building.
The AI that already knows your project.▍
every session starts from scratch.
You open an AI session on a project you've been building for months. The AI has no idea who you are, what you've built, or what matters. It asks the same basic questions it asked last week. You paste the same context block you always paste. Explain the same constraints. And the conversation that should start with solving your problem starts with onboarding the AI — again.
the "why" behind every decision disappears.
Someone removed a library three months ago. You don't remember why. The code doesn't say. The commit message says 'cleanup'. An AI working on the same feature today will confidently suggest that exact library again — because it has no idea it was a deliberate choice. That's not just annoying. It's how hours get wasted rediscovering decisions that were already made.
nothing you teach the AI ever sticks.
Every correction you make after hitting GO is gone by the next session. Every constraint you explain, every pattern you establish, every 'don't do it that way' — erased. You're not building shared understanding with your AI. You're filling a bucket that empties every time you close the tab. And the AI keeps suggesting the same things you keep rejecting.
then the AI
remembered.
the AI walks in already knowing your project.
After a one-time setup, every AI session starts with your full project knowledge already loaded — silently, before the first word. Your tech stack, your team's rules, recent decisions, things that are off-limits, and unresolved questions. The AI doesn't ask what framework you're using. It doesn't suggest what you've already rejected. It operates as if it's been on your team for months.
every important decision is saved and searchable.
When a session ends, prepcli quietly records what was done, why that approach was chosen, what was tried and rejected, and what constraints were in play. That record travels with your codebase — anyone who clones the repo can see the full reasoning, linked to the exact code change it produced. Six months from now, when someone asks 'why is it built this way?' — the answer is already there.
a new teammate hits the ground running.
Your project's full history — every decision, every rule, every lesson learned — travels with the repository. When a new developer joins, they run two commands and immediately have the same context the rest of the team built over months. The AI they work with is already calibrated to your codebase. No ramp-up wiki. No 'ask the senior dev.' No re-learning what was already figured out.
Six structured workflows installed into your AI tool. Each starts with your project context already loaded and ends with a decision record saved automatically.
- prepcli installdrops all six commands into Claude Code, Cursor, Antigravity, and Codex
- prepcli auth loginsign in with just your email — no passwords, no third-party accounts
- prepcli initreads your project once, remembers it forever
- prepcli contextsee exactly what the AI knows about your project right now
- prepcli logbrowse every AI decision linked to the commit where it happened
- prepcli session addrecords each AI turn as it happens — works offline, no login needed
- prepcli recordsave any decision to the history yourself, any time
- 6 slash commands/debug /plan /prep /refactor /review /write
- quality trackingsee how many corrections you had to make after saying GO — and watch it drop
- smarter questionsquestions that learn from your sessions and stop missing the same things twice
- session analyticswhich workflows are improving, which gaps keep coming up, and how it trends
- team sharingshared rules and history across the whole team — not just your own sessions
- setup checkone command that tells you exactly what is not configured and how to fix it
- before vs after reportshow much back-and-forth you had 3 months ago vs today
start building
with memory.
Works with Claude Code, Cursor, Antigravity, and Codex. Two commands to set up — from the next session, the AI already knows your project.