agentleFS
Sign inStart free
dmminga1983dmminga1983/minga-replica/skills/replica-architect/SKILL.md

replica-architect

Skill1 starsChanged 18 days agoMITView on GitHub

Plan the stack, database schema and API for an app clone from the recon map, for the region it will be sold in: Russia (data stored in Russia, YooKassa or CloudPayments with 54-FZ receipts, VK ID or Yandex ID), global (Stripe or Paddle, Supabase, Vercel) or both. Use for 'plan the clone', 'what stack', 'design the database', 'write the schema', 'plan the API'; RU: архитектура, какой стек, схема базы, спроектируй API.

2 other projects use a version of this, 2 of them edited.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

3 rules. About 0 are common, 1 is distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 2

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    No recon map: stop and run /replica-recon.

    line 15
  2. Very specific:

    Read replica/region.txt (written by /replica-recon): ru, global or both.

    line 19
  3. Very specific:

    Use what the user already knows.

    line 25

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

replica-architect

Reads replica/recon.md, replica/features.csv and replica/region.txt. Writes replica/architecture.md (template in this folder) and replica/schema.sql.

python3 stack.py --region ru        # or global, or both

No recon map: stop and run /replica-recon. Planning from memory misses half the app.

Region

Read replica/region.txt (written by /replica-recon): ru, global or both. It decides the stack, the stores, the laws and the trademark checks. python3 ../replica-architect/stack.py --region <region> prints them. No file yet: ask where the app will be sold and write it.

Step 1: the stack

Use what the user already knows. Otherwise take the defaults stack.py prints for the region, each with one line of why. One database. No microservices: the clone needs the original's features, not its architecture.

  • ru: database and every store of personal data (form answers, CRM, email lists, analytics with user ids) on servers in Russia. Since 1 July 2025 recording or storing Russian users' personal data in foreign databases is prohibited, including at collection. Payments through a Russian provider with fiscal receipts.
  • global: managed and cheap at zero users: Postgres on Supabase or Neon, Vercel, Stripe or a merchant of record (Paddle, Lemon Squeezy) if the user does not want to handle sales tax.
  • both: decide now where Russian users' data lives (a separate deployment in Russia, or a separate region of the same app) and which payment provider serves which cards. Write it in the "Where personal data lives" section.

Step 2: the schema

For every table: id uuid primary key default gen_random_uuid(), created_at, updated_at; an owner column on everything a user owns; foreign keys with a decided on delete; indexes on every foreign key and every filter or sort column; enums or check constraints for statuses; times as timestamptz in UTC; money as integer minor units plus a currency column; access rules (row level security, or one authorisation check per query in the data layer: write which). Hard constraints the recon found (two people booking one slot) belong in the database, not the UI.

Step 3: the API

One table per flow: method path | does | who can call | input | output | flow. Webhooks in and out, background jobs with schedules. Official public APIs with the user's own keys only, never the original's private endpoints.

Step 4: the parts that bite

One line each where it applies: time zones and daylight saving, idempotency (webhooks arrive twice), races, rate limits, uploads, search, realtime, offline, email deliverability, tenancy, account deletion, data export.

Step 5: build order

The vertical slice first (the core loop end to end, ugly), then must-haves by area, then the rest, then the fixes from /replica-users.

Output: the files, the stack in one line, table and route counts, the three riskiest parts. Next: /replica-design.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

Check your setup, free