Everything here is written against the real product and the real problem: software is being built faster than it can be trusted, and the moment it changes hands is where that gap bites. Start with the guide that matches where you are standing.
What the term means, why AI-built projects fail at the handover step, and a delivery process that survives contact with a real recipient.
Every item a handover of an AI-built project needs to cover, with a free template you can copy, and no email address required for any of it.
What an audit of AI-generated code actually checks, what human audit services charge, and what executed verification proves for £29.
A plain-English guide for the person receiving the project: what to ask for, what the labels mean, and the one command that checks everything.
Rescue engagements start at four figures. Run a verification first, know what is actually broken, and brief the developer with evidence.
A repeatable delivery standard for teams shipping AI-assisted client work: what to hand over, in what form, and how to prove it.
A short definition of the artifact: a verified, evidence-labelled record of what a software project contains, generated at handover.
What the Act changed for teams delivering AI-assisted software, what a provenance record contains, and how to generate one free with one command.
The undocumented SQLite layout behind Cursor's chat history, how to find and read it on every platform, and the traps waiting inside it.
Claude Code keeps a full transcript of every session as JSONL files on your machine. Here is exactly where they live and how to read them.
Codex CLI writes a rollout file for every session. Where they live, how sessions map back to projects, and why the ChatGPT app has no equivalent.