The read-and-approve layer for codebases in the AI era.
Do not send AI into your codebase blind.
DevReadr opens your repository like a workbench: clickable files, code on one side, plain-English intent on the other, and a Readr Twin that tracks meaning, marks, risk, tests, and approval.
AI code is fast. Production code needs position, shared meaning, tests, and approval.
This checkout file starts the payment flow.
The accepted meaning is stricter: a redirect is not payment proof. Tenant activation should wait for Stripe webhook confirmation.
Meaning locked. Watch future changes for drift.
export async function createCheckoutSession(input) {
// Redirects are theater. Webhooks are truth.
return stripe.checkout.sessions.create({
mode: 'subscription',
metadata: {
tenantId: input.tenantId,
activateAfter: 'invoice.paid'
}
});
}Every codebase gets a Readr Twin.
A Readr Twin gives humans and approved agents a shared map of meaning, risk, tests, and approval before anyone drafts changes.
Explore
Click through the live repository like an IDE, with folders, files, maps, and ownership in view.
Translate
See selected code beside plain-English intent, risks, dependencies, and suggested tests.
Mark
Lock meaning, flag a human, mark AI-ready context, or watch for code-to-meaning drift.
Draft
Turn corrected intent into Patchbit plans, tests, rollback notes, and human-approved next actions.
The repo is no longer just files.
The repo becomes code plus meaning. That is the layer teams need when engineers, leaders, clients, and AI agents are all touching the same system.
Meaning lock
Accepted plain-English intent becomes a reference point for future changes.
Team marks
Engineers, founders, PMs, operators, and clients can flag what matters without losing code position.
Drift watch
When code moves away from accepted meaning, DevReadr should make that visible before it becomes a bug.
Approved agent context
DevReadr gathers position, intent, risks, tests, and approval state before any approved agent drafts.
Built for serious teams using AI.
DevReadr is not another docs site and not an unguarded agent. It is the working surface between code, people, and approved action.
Engineers move faster because the repo, code, intent, risks, and tests sit in one workbench.
Non-engineering leaders can understand what the code does without pretending code review is optional.
AI agents stop working blind because every task starts from scoped context and a known definition of good.
Teams get a safer approval layer for code review, onboarding, migrations, PR planning, and client scope.
Private by default.
DevReadr should help customers move faster without pretending AI output is automatically safe.
Customer-controlled access
Customers choose which repositories, schemas, or files enter DevReadr. Access can be removed by request.
Derived context by default
Starter onboarding should retain maps, notes, marks, and metadata instead of source material.
No exposed tokens
Tokens must stay out of client code, public pages, screenshots, and committed files.
Draft-only unless approved
Patchbit plans, PRs, migrations, deploys, and external actions require human approval unless a scoped auto-policy exists.
Hook up your codebase, run a review, and get started.
Founding beta is for the first teams that want a selected codebase turned into a Readr Twin with maps, plain-English intent, team marks, and draft-only Patchbit planning.
$28 first month
Private beta checkout for first cohort.
Target under $100/month
Multiple codebases for founders and small teams.
Scoped manually
Collaboration, permission review, and guided rollout.
Build your first Readr Twin.
Sign in creates identity. Checkout activates the paid workspace. GitHub repo access connects inside the workbench.
Prefer to collaborate first? Email hello@devreadr.com.