Guide · Agent skill

How to use the Replica Architect agent skill

Updated 2026-11-02 · written with AI from the project's own documentation and checked against our catalogue data

Replica Architect is an agent skill: a folder of instructions that an AI coding agent such as Claude loads when a request matches its description. Once it is installed, you can ask the agent to plan the stack, database schema and API for a clone of an app, and it will write an architecture document. This guide shows how to install it, how to trigger it, what it needs before it runs, and where to be careful.

Replica ArchitectJakeschincariol/replica-skill · MIT Download the skill

Quick start

1Get the skill files. Download the ZIP from this page and unzip it to get SKILL.md and architecture.md.
2Place the folder. Unzip into ~/.claude/skills/ for all projects, or into the project's .claude/skills/ folder for one project. Alternatively, upload it under Settings → Capabilities → Skills in the Claude apps.
3Run replica-recon. Produce replica/recon.md and replica/features.csv with the replica-recon skill. The architect skill cannot plan without them.
4Ask for the plan. Make a natural request such as "plan the clone" or "design the database" in your project.
5Review architecture.md. Check the stack, schema, API table and build order. Confirm the access rules and deletion handling.
6Build the vertical slice first. Start with the thin end-to-end slice the plan lists, then move to must-haves in order.

What the skill makes the agent do

The skill reads a feature map produced by the companion skill replica-recon, specifically replica/recon.md and replica/features.csv, and writes replica/architecture.md. It works through five steps: choosing a stack, turning the inferred data model into SQL, listing every route by user flow, noting the tricky parts, and ordering the build.

Stack: a default modern set of managed tools, with one line of reasoning per choice.
Schema: SQL tables with ids, timestamps, owner columns, foreign keys, indexes and access rules.
API: every route or server action, with who can call it and which flow it serves.
Build order: a thin end-to-end slice first, then must-haves, should-haves and could-haves.

Install the replica-architect skill

Download the ZIP from this page and unzip it. The package contains SKILL.md and architecture.md. Put the folder in one of two places, depending on whether you want the skill for all projects or one project.

You can also upload the skill in the Claude apps under Settings → Capabilities → Skills.

All projects: unzip into ~/.claude/skills/ so the skill is available everywhere.
One project: unzip into the project's .claude/skills/ folder.
Claude apps: upload the skill under Settings → Capabilities → Skills.

Run replica-recon first

The skill needs a recon map. It reads replica/recon.md and replica/features.csv, which are produced by the replica-recon skill and are not included in this package. If those files do not exist, the skill stops and asks you to run /replica-recon first. This is deliberate: planning a clone from memory of what an app does tends to miss features.

Trigger the skill with natural requests

You do not need a special command. The skill's description lists phrases that should activate it. Describe what you want in plain language, for example:

"Plan the clone"
"What stack should I use?"
"Design the database"
"Write the schema"
"Plan the API"
"Architecture for my version of X"
Or run it straight after /replica-recon finishes.

Example: a booking app

The source's example is a booking app. For its core table, the skill produces a bookings table with a UUID primary key, a foreign key to event types and hosts with an on delete rule, start and end times stored as timestamptz, guest timezone, a status field limited by a check constraint, and a constraint that the end time is after the start time. It adds an index on host and start time. Two guests booking the same slot is handled in the database, with an exclusion constraint or unique index, not only in the interface.

create table bookings (
  id uuid primary key default gen_random_uuid(),
  event_type_id uuid not null references event_types(id) on delete cascade,
  host_id uuid not null references users(id) on delete cascade,
  start_at timestamptz not null,
  end_at timestamptz not null,
  guest_name text not null,
  guest_email text not null,
  guest_timezone text not null,
  status text not null default 'confirmed'
    check (status in ('confirmed','cancelled','rescheduled')),
  answers jsonb not null default '{}',
  created_at timestamptz not null default now(),
  constraint no_zero_length check (end_at > start_at)
);
create index on bookings (host_id, start_at);

Review the output and the safety limits

The skill declares no scripts and makes no network calls of its own. Its instructions limit it to official, public APIs using the user's own keys, and forbid using the original app's private endpoints, even if they are visible in a browser. The default services it names (such as Next.js, Supabase or Neon, Stripe and Vercel) are suggestions; you only need accounts for the ones you choose.

Read the generated SQL and architecture before you build. The skill's output is a plan, and the database rules, deletion handling and payment flows need a person who understands them.

Scripts: none declared.
Network: none declared.
Accounts: only for services you choose.
Cloning for commercial imitation carries legal and terms-of-service risk.

Tips for better results

Give the skill a clear picture of your own version before it plans. Say which stack you already know, since the skill prefers your existing choices over its defaults. Keep the scope to the core features in the recon map.

Name your existing stack in the request to skip the defaults.
Ask for the riskiest parts summary if you want a short list to review first.
Use the output as input for the next step, /replica-design, once architecture.md is written.

What people use it for

Planning a from-scratch rebuild of a web or mobile app's core features.
Getting a concrete stack, schema and build order before writing code as a solo founder.
Producing a checklist of tricky parts such as time zones, webhooks, race conditions and GDPR deletion.
Drafting a route list by user flow for a team to review.

Frequently asked questions

How do I install the replica-architect skill?

Download the ZIP from this page and unzip it. Put the folder in ~/.claude/skills/ for all projects, or in a project's .claude/skills/ folder for one project. You can also upload it in the Claude apps under Settings → Capabilities → Skills.

Do I need replica-recon to use replica-architect?

Yes. The skill reads replica/recon.md and replica/features.csv, which replica-recon produces. Those files are not included in this package, and without them the skill stops and asks you to run /replica-recon first.

Is replica-architect free and what is its licence?

The skill is published under the MIT licence. The source does not state prices for the services it suggests; accounts are needed only for the services you choose to use.

Does the skill run scripts or use the network?

According to its listing, it runs no scripts and makes no network requests. The third-party services it names are optional suggestions that you would set up yourself.

What stack does replica-architect recommend?

Its default is Next.js with TypeScript, Postgres on Supabase or Neon, Drizzle or Prisma, Supabase Auth or Auth.js, Stripe, Resend or Postmark, Vercel for hosting, and Expo for mobile. If you already have a stack, the skill uses that instead, and each choice comes with a one-line reason.

How does it compare to planning the architecture by hand?

The skill works from a feature map and returns a structured plan with SQL, a route table and a build order in a fixed format. It does not replace reviewing the output, and the source says it refuses to plan from memory of an app.

New agent skills, repos and AI tools — every week

The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.

Our verdict

Replica Architect suits developers and solo founders who already have replica-recon output and want a concrete stack, schema and build order for a from-scratch rebuild. It is a poor fit for anyone without that recon map, for cloning an app for commercial imitation or using private endpoints, and for non-technical users who cannot review the generated SQL.

Download the skill