Before you build

Set the boundaries before your AI builds.

Describe your project. Decide who can do what, which limits apply, and what happens when something fails. Get a clear brief to give your coding agent.

See how it works Not sure how to describe it?

Already have a working project? Review it and plan the next improvements.

Try an example

Start with one click.

Each example starts a browser-local review. Nothing is compiled until you choose to create the files.

What are you asking your AI to build?

Keep it high-level. Do not paste secrets, customer records, credentials, or source code. We check the local library first, then show you what matched.

Optional settingsContext about you, and a Cloudflare AI fallback

About you (optional)

Choose the setting closest to your work. It does not affect the review and is counted only as a group.

Use Cloudflare AI if the local check finds nothingOff by default. If we find no matches, your description is sent once to Cloudflare AI for possible suggestions. We do not store the description or response, and you approve every suggestion. Optional

No account or repository access. Matching runs in your browser. Creating the files sends your description and answers to our Cloudflare-hosted compiler; they are not stored. The Cloudflare AI check is separate and optional. Do not include secrets or private data.

A clearer hand-off

Three useful things to settle first.

See what matters
Find the parts of your idea that affect how it should be built.
Make the important calls
Set limits, approvals, and scope—or leave them open for later.
Hand over a useful brief
Give your agent requirements and a way to report evidence, with recipes and validation scenarios where available.

What you get

Ready for your agent

Build instructions.agentdirective/agent-directive.jsonStructured rules your coding agent can follow.
Readable project brief.agentdirective/AGENTDIRECTIVE.mdA plain-language version for people and agents.
Implementation progress.agentdirective/control-status.jsonWhat your agent reports as built, incomplete, or still open.

Clear instructions, not a vague score.

Your agent gets the boundaries to implement, available validation scenarios, and a structured way to report progress.

How it works

From an idea to a clear hand-off.

Six steps, in order. Scroll to move through them.

01 / Describe
Tell us what you want to build.
No account, repository, or source code. Just the description you were going to give your coding agent.
What are you asking your AI to build?
Build a support tool that reads customer emails and automatically issues refunds.
Review my build Local matching; optional AI is separate.
02 / Detect
We identify what the system does.
Published rules look for known traits. You see the list and can correct it before continuing.
Traits found in your description
AI componentOutside inputMoney movementCustomer data
Every trait is yours to correct or remove.
03 / Match
Your confirmed traits select the safeguards.
Published rules decide what applies, and each pack explains why it was selected.
Safeguards that apply
- AUT-001
  Bounded AutonomyBecause money can move without a person.
- AI-001
  Untrusted ContentBecause a model reads text you did not write.
- OBS-001
  Traceable ActionsBecause someone will have to answer for a refund.
04 / Decide
You answer the questions that matter.
Limits, approvals, and scope stay open until you confirm them. Nothing invents a business value for you.
AUT-001-Q1 · Threshold
What is the largest refund an automation may issue without human approval?
$500Any amountLeave open — a valid answer
An open decision is recorded as open, and your agent is told to ask.
05 / Hand off
Your agent gets clear instructions.
A directive it can implement, plus the files it uses to report what it built — and what it could not.
.agentdirective/
- agent-directive.jsonThe boundaries, as structured rules.
- AGENTDIRECTIVE.mdThe same thing, for a person to read.
- control-status.jsonWhere your agent reports what it built.
06 / Report back
Your agent records what it built.
It reports what it implemented, what it could not, and the evidence behind each claim. Reassess the same project later against its directive.
control-status.json
- implemented
  APPROVAL_THRESHOLD2 tests named, evidence pointed at.
- partial
  ACTION_TRAILLimits and remaining work recorded.
- unresolved
  EXPORT_SCOPEStill yours to decide.
Reported is not verified, and the report says so.

See the full walkthrough →

Why it helps

Make the important decisions yourself.

A coding agent may make assumptions when a prompt leaves something out. Review the choices that affect authority, data, and failure handling before implementation.

Your decisions stay yours

Confirm the limits, approvals, and scope you know. Leave the rest open. The directive tells your agent to ask before choosing those values.

See why a safeguard applies

Each matched Control Pack links a project trait to a potential failure, an implementation requirement, and supporting sources.

Give your agent useful instructions

The brief includes requirements, tests to run, and a structured way to record progress. Packs with implementation recipes include those too.

Check the reported evidence

An optional follow-up finds missing or inconsistent evidence in your agent’s report. It does not inspect source code or test the running system.

Where to start

Start where you are.

No account, no repository access, no source code. Pick whichever matches where you are.

Describe a build Paste the prompt you were about to give your coding agent and review it before anything is built. Go to the description box Pick what you are building Choose a project type and its main actions. Then review and correct the suggested traits. Open the guided start Review a project you already have Create an inspection prompt for your agent, or bring a previous directive to review what changed. Review an existing project

What this is not

Design guidance, not a security guarantee.

AgentDirective prepares instructions and checks reported metadata. You still need code review and testing to find out whether the implementation follows those instructions.

Your review

First, let’s confirm what we understood.

Your journey

Step 2 of 5 · Confirm

1
Describe
2
Confirm
3
Decide
4
Download
5
Improve

01 / Check the result

These traits may affect how the system should be built.

You are in control. Remove anything that does not apply, or leave it open for the next step.

Why these safeguards matched

We show how your confirmed traits connect to each safeguard before you decide.

How much should we review?

Quick Review focuses on the most material unresolved decisions.

Local matching runs first. Cloudflare AI is used only after zero local matches and explicit consent.

04 / Ready to use

Your Agent Directive is ready.

Give these files to your coding agent as its design boundaries and reporting rules.

Next: give your coding agent the instructions

Copy a complete, self-contained prompt with your confirmed boundaries, open decisions, and reporting rules. Paste it into your coding agent. Copying sends nothing new; the text comes from the compiled result in this browser.

1. Hand offPaste the prompt or attach the complete ZIP bundle.
2. Build & reportYour agent asks about open decisions, then records its work and evidence.
3. Check the claimsReturn with the directive and status files for an optional metadata review. Review code and tests separately.

Your design at a glance

How the pieces connect

A suggested flow based on your analysis—not a scan of your code. Confirm the components, directions, and return paths. Select any watch-out for safeguards, tests, and evidence. Missing warnings do not establish safety; your directive’s decisions remain the source of truth.

Generated in this browser. Opening the map uses a one-time tab handoff that is deleted when read.

Did this review change your design?

Optional. We store only one aggregate category count—never your description, directive, identity, project, or free text.

What changed?

Ask your agent for a report when it finishes

The agent prompt above already includes this step. Copy it on its own if you have already handed off. Your agent runs one command and writes an implementation report into the project: every requirement, the value you decided, what it reported building, and what it built without asking. The report is generated from the status file, not written by the agent.

What's actually in these files?

Applied safeguards

Each pack below shows what it covers, the decision recorded, the evidence expected, and the most relevant implementation shape for your coding agent.

Still unresolved

What happens next

Your agent can continue from here.

Give the bundle to your coding agent. It can read the directive, report implementation evidence, and record meaningful changes for the next review.

ReadThe agent reads the directive and implementation plan.
ReportStatus, tests, and limitations are logged.
ReassessMaterial changes start the loop again.

05 / Improve · Optional follow-up

Review what your agent reported.

Once your coding agent writes its status file, submit it with your directive to check the reported evidence for missing or inconsistent claims. This does not verify the source code or runtime behavior.

Review reported evidence

Design guidance, not a security guarantee

This directive gives your agent explicit design boundaries to follow. It provides clear instructions and reporting requirements, but does not replace code review or runtime testing.

Set the boundaries before your AI builds.

From an idea to a clear hand-off.

Tell us what you want to build.

We identify what the system does.

Your confirmed traits select the safeguards.

You answer the questions that matter.

Your agent gets clear instructions.

Your agent records what it built.