EmbedIQ

<!-- audience: public -->

# EmbedIQ

**Governed configuration for every AI coding agent — from one interview.**

*Answer a few questions about your project, and EmbedIQ writes the setup files
that make Claude Code, Cursor, Copilot, and other AI coding assistants follow
your team's rules — consistently, and with an audit trail.*

[![License: MIT](https://img.shields.io/badge/License-MIT-blue.svg)](LICENSE) · [Latest release: **v4.0.5**](https://github.com/asq-sheriff/embediq/releases/latest) · Deterministic & offline

EmbedIQ interviews you once about your project, team, and compliance
obligations, then generates a complete, governed agent harness — typically
15–40 files — for every major AI coding agent from a single answer set:
**Claude Code, Cursor, GitHub Copilot, Gemini, Windsurf, and cross-agent
`AGENTS.md`**. The generator makes **zero LLM calls**, so the same answers
always produce **byte-identical output** — including under regulatory audit.

[Full changelog](CHANGELOG.md) ·
[Security model](SECURITY.md) ·
[Documentation](docs/getting-started.md)

---

## What EmbedIQ does, in plain terms

AI coding assistants — Claude Code, Cursor, GitHub Copilot, and the rest — each
read a small **setup file** that tells them the rules of *your* project: what
you're building, what language it's written in, and what they must never do
(at a hospital, for example, *"never put a patient's record into code, comments,
or logs"*). Writing those files by hand for every tool — and keeping them all
saying the same thing — is slow and easy to get wrong.

**EmbedIQ does it for you.** You answer a set of plain questions about your
project, your team, and any rules you have to follow. EmbedIQ turns those answers
into all of the setup files automatically — the same way every time, with no AI
guesswork in the middle.

- **Who it's for** — any team using AI coding assistants that wants them set up
  the same way for everyone, especially in regulated fields like healthcare,
  finance, or government, where an assistant doing the wrong thing is a
  compliance problem, not just a bug.
- **What you do** — answer an interview (10–15 minutes), in the browser or the
  terminal. No coding required to run it.
- **What you get** — a folder of ready-to-use config files you drop into your
  project, so every developer's AI assistant follows the same safe rules from
  day one, plus a record you can show an auditor proving what was set up and why.

---

## See it in action

One interview, four steps — from a role-scoped handoff to a folder of ready-to-commit files.

**1. Start the role-scoped handoff.** Roles unlock in order — the Admin sets security & policy, then hands off to the Team Lead (project & tech) and the Individual (per-seat preferences). The Team Lead and Individual stay locked until the prior slice is done.

![EmbedIQ welcome screen with the Admin → Team Lead → Individual handoff: the Admin step active, Team Lead and Individual locked](docs/assets/flow-welcome.png)

**2. Answer only your slice.** Each role is asked just the questions it owns — seven dimensions in the sidebar, role-adaptive wording, and an admin-only "why we ask" panel.

![The EmbedIQ interview scoped to the Admin slice — dimensions sidebar, a role question with help text and the "why we ask" panel](docs/assets/wizard.png)

**3. Review and hand off.** Each slice ends in a summary of what's configured so far, then passes the baton to the next role.

![The "Admin slice complete" summary with the handoff stepper and a profile/priorities recap, ready to continue as the Team Lead](docs/assets/flow-handoff.png)

**4. Files ready — pick a location and generate.** After the final slice, EmbedIQ shows every file it will write and asks where to put them.

![The "Ready to Generate" screen: a target-directory input and the full list of files to generate](docs/assets/flow-generate.png)

---

## Why it matters

Most teams use four to six AI coding tools at once, and each one wants its rules
written in its own way. So the same rules get copied from tool to tool, fall out
of step as people edit them by hand, and quietly go stale. Compliance teams have
no single place to check; every developer's machine ends up a little different;
and each new hire sets the whole thing up again from scratch.

EmbedIQ replaces all of that with **one source of truth** built from your
answers — and keeps it that way. It spots when files have been changed or have
drifted out of date, can regenerate them on a schedule, and produces the exact
same output every time. Because no AI runs while the files are written, the
result is repeatable and provable, not a best guess.

---

## What the compliance standards mean — and why enterprises care

In a large company, much of EmbedIQ's value is keeping AI assistants on the
right side of rules the company is legally or contractually bound to. Those
rules have names. Here's what they are in plain English, and what's at stake
when they're broken:

| Standard | What

... (truncated)