AI Skill StudioAI Skill Studio
Skill Library
Presets:

Target Standard & File Format

AI Agent Rules & Skills
Context Shields & Privacy (Ignore Files)
Integrations & Machine Documentation
AI Governance & Multi-Agent Planning Suite

Identity & Activation Rules

3 chips
Scoped domain keywords evaluated in real time
api-routescode-auditsrefactoring
Suggested Chips:

Technology Stack Context

Engineering Philosophy

Flexible & Adaptable

Architectural & Code Quality Conventions

Less rigid & configurable

Agent Behavioral Guardrails

Select active rules
Security & Trust Boundaries
Always included
  • Untrusted Content Is Data
  • Encoded Payloads
  • Least-Privilege File Access
  • Confirm Irreversible Actions
  • Secret Handling
  • Validation Failures
  • Conflicting Requests

Step-by-Step Workflow Procedures

Backend · Any coding agent

Bun & Elysia / Hono ARCHITECTURE.md

An ARCHITECTURE.md for Bun & Elysia / Hono: module responsibilities, data flow and the boundaries code changes must keep.

Path
ARCHITECTURE.md
Loads
Read when referenced
Format
Markdown (.md)
Used by
Any coding agent
All ARCHITECTURE.md presets and how ARCHITECTURE.md works

What this file tells the agent

ARCHITECTURE.md describes module responsibilities, state boundaries and data flow, so a coding agent knows where code belongs before it changes anything. Point to it from CLAUDE.md or AGENTS.md so agents read it.

  • Use Elysia's native t schema builder (powered by TypeBox) for request body, query, and parameter validation.
  • Leverage native Bun.serve(), Bun.file(), and Bun.password instead of slow Node.js polyfills or external dependencies.
  • Structure plugins using .use() and define explicit Eden Treaty contracts for frontend consumption.
  • Handle application errors using Elysia .onError(({ code, error, set }) => ...) with typed error codes.
  • Ensure zero unhandled promise rejections and use strict type inference for all endpoint responses.

Code to avoid, and what to write instead

The rule steers the agent away from the left and toward the right.

avoid
// Discouraged: Node.js bcrypt imports, unvalidated body, and Express paradigms
import bcrypt from "bcrypt";
app.post("/users", async (req: any, res: any) => {
  const hash = await bcrypt.hash(req.body.password, 10);
  res.json({ success: true });
});
write instead
import { Elysia, t } from "elysia";

export const app = new Elysia()
  .onError(({ code, set }) => {
    if (code === "NOT_FOUND") {
      set.status = 404;
      return { ok: false, error: "Resource not found" };
    }
  })
  .post(
    "/api/users",
    async ({ body, set }) => {
      const hash = await Bun.password.hash(body.password);
      set.status = 201;
      return { ok: true, email: body.email };
    },
    {
      body: t.Object({
        email: t.String({ format: "email" }),
        password: t.String({ minLength: 8 }),
      }),
    }
  );

Install it in one command

  1. 01

    Go to your project root

    The file goes at ARCHITECTURE.md, relative to the root.

  2. 02

    Download the file

    Run the command on the right. It downloads this preset and writes the file.

  3. 03

    Make sure it gets used

    Agents read it when your CLAUDE.md or AGENTS.md points to it. Add a line such as "Read ARCHITECTURE.md before structural changes".

$ curl -fsSL "https://www.devscratchpad.tech/api/raw/architecture-md/bun-elysia" -o "ARCHITECTURE.md"

writes ARCHITECTURE.md

path

ARCHITECTURE.md

To write the file from your own code instead, use Read your code or Convert Legacy Rules in the editor above.

Questions

What belongs in ARCHITECTURE.md?
How the main modules relate, how state and data flow, and the boundaries every change must keep.
How does ARCHITECTURE.md help AI agents?
It tells them where code belongs and which boundaries to keep, so they are less likely to add circular dependencies or skip layers. Point to it from CLAUDE.md or AGENTS.md so they read it.