agentleFS
aya-talebaya-taleb/mimic-skills/skills/mimic-backend/SKILL.md

mimic-backend

Skill1 starsChanged 22 days agoMITView on GitHub

Wires up the parts that make your version real and safe: sign in, payments, email, and the integrations the recon found, all through official APIs and SDKs with your own accounts and keys. Secrets in env vars, signed and idempotent webhooks, billing complaints from reviews fixed in the product, then a security checklist marked pass, fail or not checked. Use when the user says "add login", "set up auth", "add Stripe", "take payments", "send emails", "connect Google Calendar", "add the integrations", "backend", "is it secure", "security check", or after /mimic-build. Also covers payment links, booking tools and email lists for services and local businesses.

  • Reads credentials

The file

We split this skill into the rules it gives an agent

8 rules. About 0 are common, 6 are distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈6

    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:

    Card data never touches the server: Use the payment provider's hosted checkout or its own embedded fields.

    line 40
  2. Distinctive:

    Passwords: use the auth provider, or a slow hash (argon2id or bcrypt) if the stack rolls its own.

    line 42
  3. Distinctive:

    Costs: if a service is paid beyond a free tier, say so, link its pricing page, and wait for the user's yes before they sign up for a paid …

    line 45
  4. Distinctive:

    Fix the billing complaints in the product: If reviews said "hard to cancel" or "surprise renewal", build the fix here: cancel in one click …

    line 70
  5. Very specific:

    If a feature needs the competitor's own data or network, it cannot be integrated.

    line 92
  6. Very specific:

    Write mimic/backend.md on every run, even a short one, with the checklist as it stands.

    line 112
  7. Distinctive:

    Every file and reply is read by a busy founder.

    line 135
  8. Distinctive:

    If the humanize skill is installed, run copy that a person will publish or send through it.

    line 138

≈ “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

mimic-backend

This is where a demo becomes something people trust with their money and data. Most of the damage a young product can do happens here, so take each piece slowly and check it.

Reads and writes

Reads: mimic/architecture.md (stack, services, schema), mimic/features.csv (which integrations are must), mimic/fixes.md (billing and trust complaints), the code from mimic-build, and the business context.

Writes:

  • the code for auth, payments, email and integrations
  • .env.example with variable names only, never values
  • mimic/backend.md (template: references/backend-template.md): each service, why, the account and keys the user creates, env var names, webhooks and how they are verified, test mode steps, and the security checklist

The rules

  • Official routes only. Each service through its official API or SDK, with an account the user creates and keys the user pastes into their own env. Never call a competitor's private API, reuse someone else's keys, or log into a platform as the user.
  • Secrets never touch the repo. Env vars only. .env is in .gitignore. Ask the user to paste keys into .env themselves rather than into the chat.
  • Card data never touches the server. Use the payment provider's hosted checkout or its own embedded fields. Never store card numbers. If asked to, decline and explain.
  • Passwords: use the auth provider, or a slow hash (argon2id or bcrypt) if the stack rolls its own. Never plain text, never reversible.
  • Nothing is marked passed without a check. "Looks fine" is not checked.
  • Costs: if a service is paid beyond a free tier, say so, link its pricing page, and wait for the user's yes before they sign up for a paid plan.

Step 1: auth

Pick from the architecture. A hosted auth provider or the framework's well known auth library is the default; rolling your own is rarely worth it.

  • sign up, sign in, sign out, password reset or magic link, email verification
  • sessions: http-only, secure, same-site cookies; sensible expiry; sign out everywhere
  • social sign-in only for providers the users actually use, with the smallest scopes
  • access checks on every record: a user can only read and write their own rows. Put the rule in the database (row level security) or in one shared check every route uses. This is the bug that leaks other people's data, so test it on purpose.

Step 2: payments

  • products and prices created in the provider by the user (list them from mimic/launch/pricing.md if it exists, else from the architecture)
  • hosted checkout or the provider's elements, in test mode first
  • a webhook endpoint that verifies the signature with the provider's SDK, handles each event once (store the event id and skip repeats), and is the only thing that grants or removes access
  • the customer portal or an equivalent for cancel and update card

Fix the billing complaints in the product. If reviews said "hard to cancel" or "surprise renewal", build the fix here: cancel in one click from settings, an email before renewal, receipts that show what was charged. Note each fix with its evidence.

Step 3: email

  • a transactional email provider for sign-in, receipts, reminders and notices
  • the domain records the provider gives (SPF, DKIM, and a DMARC policy). List them for the user to add at their registrar. mimic-deploy checks them.
  • any email that is not transactional (newsletters, promotions) needs consent and a working unsubscribe. Marketing to people who did not opt in is not something this skill sets up.

Step 4: integrations

For each integration in features.csv (calendar, maps, messaging, storage...):

  • the official API, the auth method (OAuth with the smallest scopes that work, or an API key the user creates), and the webhooks or polling it needs
  • what the product does when the integration fails or a token expires
  • rate limits from the provider's docs, linked

If a feature needs the competitor's own data or network, it cannot be integrated. Mark it skip in features.csv with the reason.

Step 5: security checklist

Go through references/security-checklist.md and mark each item pass (with how you checked), fail (with the fix), or not checked (with why). Fix every fail you can now. Put the result in mimic/backend.md. mimic-deploy reads it, and any fail there blocks launch.

Services and local businesses

Same rules, different tools: payment links or a point of sale instead of custom checkout, a booking tool instead of a booking engine, an email list with consent, a shared sheet or small admin page for orders. Still: accounts in the user's name, keys in env vars or the tool's settings, two-factor sign-in on every admin account, and the same checklist for who can see customer data.

Always

Write mimic/backend.md on every run, even a short one, with the checklist as it stands. mimic-deploy reads it. End the reply with "Next: /mimic-test" once the pieces asked for are in place.

For every must integration in features.csv, either build it, or write in backend.md why it is not built yet. A check marked pass names the test or command that proved it, and the check has to cover the whole item (a search for one key prefix does not prove "no secrets in code"). Reading the code is "reviewed, not tested", which is not checked. "Nothing exists yet" and "by design" are not checked, not pass.

Package versions, plan limits and vendor rules you did not read in this session are written as "check:" with the docs link, not as facts.

Edge cases

  • No keys yet: build against test mode, leave the env var names in .env.example, and give the steps to get each key.
  • "Just hardcode the key for now": decline. Put it in .env and show how.
  • Asked to scrape a platform or automate a user's social accounts: decline. Offer the official API if one exists, or a manual step.

Writing

Every file and reply is read by a busy founder. Plain words, sentences under 25 words, no em dashes, no hype. Skip these words in everything you write: delve, leverage, seamless, streamline, comprehensive, journey, unlock, elevate, empower, foster, robust, crucial, utilize, simple. <!-- allow-words --> If the humanize skill is installed, run copy that a person will publish or send through it.

Output

Code, .env.example, mimic/backend.md with the checklist, and a summary: what is wired, what is in test mode, what the user must create or paste, and any fail left.

Next: /mimic-test.

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.