Architecture As Code Javascript logo

Architecture As Code Javascript

Organization
l-gevity
architecture-as-code-javascript

JavaScript / TypeScript implementation of the `architecture-as-code` pattern. Per-subsystem `eslint.architecture.mjs` files merged into a single ESLint flat-config and enforced via `eslint-plugin-boundaries`. TRIGGER when: implementing or extending architecture-as-code in a JS/TS repo, debugging an `eslint-plugin-boundaries` rule, or adapting the assembler. SKIP for routine edits inside a governed subsystem. Reads in conjunction with `architecture-as-code` (the pattern, source of truth for schema, rule placement, anti-patterns, and audit checklist) — this skill defines only the JS-specific encoding, assembler code, and gotchas.

Overview

Publisherl-gevity
Repositoryl-gevity-skills
Skill namearchitecture-as-code-javascript
Stars
72
Forks
13
Bundled files
1
LicenseMIT
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • 1 bundled files

    Scripts, templates, and references the model can read while it works. Files are read-only and never executed.

  • Open source

    Published by l-gevity on GitHub. Read the source before you install it.

Installation

Install the Architecture As Code Javascript AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
    https://github.com/l-gevity/l-gevity-skills/tree/main/.agents/skills/architecture-as-code-javascript
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/l-gevity/l-gevity-skills.git /tmp/l-gevity-skills
mkdir -p .claude/skills
cp -r /tmp/l-gevity-skills/.agents/skills/architecture-as-code-javascript .claude/skills/architecture-as-code-javascript
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Architecture As Code Javascript in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Architecture As Code Javascript on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Architecture As Code Javascript is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

Architecture-as-Code — JavaScript Implementation

Prerequisite. Read architecture-as-code first. The schema (§1), subsystems (§2), forbidden edges (§3), rule placement (§4), assembler concept (§5), and anti-patterns / audit (§6) are defined there and apply identically here. This file documents only what is JavaScript-specific.

1. File format

  • Filename: eslint.architecture.mjs. Use .mjs only; .js trips source-discovery walkers and ESLint's own config-loader.
  • ES module with export default { subsystems: [...], forbidden: [...] }, plus an optional externals: [...] for npm-package policy (§3). forbidden relates declared subsystems to each other and cannot name a package. components is read as a deprecated alias for subsystems (§2).
  • Pattern syntax: filesystem globs (<dir>/**).
  • Repo-root package.json must include "type": "module".
js
// eslint.architecture.mjs — example for a subsystem with internal layering
export default {
    subsystems: [
        { name: 'core-facade', pattern: 'packages/core/index.js', mode: 'file' },
        { name: 'core-tier1',  pattern: 'packages/core/tier1/**' },
        { name: 'core-tier3',  pattern: 'packages/core/tier3/**' },
        { name: 'core-other',  pattern: 'packages/core/**' },        // catch-all, last
    ],
    forbidden: [
        // Efferent — self-knowledge, lives in own file.
        { from: 'core-*', to: '*', except_to: ['core-*'],
          why: 'Core purity: no imports outside the core directory.' },
        // Internal layering — own-prefix only.
        { from: 'core-tier3', to: 'core-tier1',
          why: 'Tier 3 must go through tier 2.' },
    ],
};

Boundaries-plugin's parametric "not equal" syntax for cross-domain isolation is !{{from.captured.domain}} (double braces).

2. Assembler

eslint.config.js discovers every eslint.architecture.mjs deeper-first at lint startup, merges and expands them, and emits one eslint-plugin-boundaries block whose files glob equals the linted source set, with the file-existence and dependency rules both at error. Read references/assembler.md before creating or changing eslint.config.js, the assembler, or that block's files, severity, or ignores: it holds the code and why each line is shaped the way it is.

3. JavaScript-specific enforcement recipes

Confine a provider SDK to its adapter

The architecture-guidelines §10 handoff usually arrives as "only the adapter may talk to the provider". A provider SDK is an npm package, not an element, so forbidden cannot express it: that list only relates declared subsystems to each other. Declare package policy separately.

js
// eslint.architecture.mjs
export default {
    subsystems: [
        { name: 'email-adapter', pattern: 'api/src/email/**' },
        { name: 'api-other',     pattern: 'api/src/**' },   // catch-all, last
    ],
    forbidden: [ /* element-to-element edges */ ],
    externals: [
        { package: '@azure/communication-email', allow: ['email-adapter'],
          why: 'The provider SDK is confined to its adapter; callers use the port.' },
    ],
};

Assembler translation. Policies are evaluated in order and the last match wins, so deny the package everywhere first, then re-allow its owning elements:

js
const allExternals = archs.flatMap(m => m.default.externals ?? []);
const externalRules = allExternals.flatMap(x => [
    { from: expand('*'),
      disallow: { to: { module: { origin: 'external', source: x.package } } },
      message: x.why },
    { from: expand(x.allow),
      allow: { to: { module: { origin: 'external', source: x.package } } } },
]);

'boundaries/dependencies': ['error', {
    default: 'allow',
    checkAllOrigins: true,                       // REQUIRED — see below
    rules: [...rules, ...externalRules],
}],

Two traps make this silently wrong:

  • Without checkAllOrigins: true the policy never fires. The rule reports nothing, the lint is green, and the boundary does not exist. The flag is not an optimization.
  • checkAllOrigins: true subjects every package to the block's default. With default: 'disallow' that turns one SDK policy into a repo-wide import ban. Keep the blanket default: 'allow' and express prohibition as rules, or add an explicit allow-all policy first.

Register the rule at error. At warn this is a report, not a boundary.

Prove it red first. Add a throwaway file that imports the package from outside the adapter, confirm the lint fails on it with the declared message and line, delete the file, then confirm the adapter and its callers pass. A policy that has never failed has not been shown to run — the checkAllOrigins trap above produces exactly the same green output as a working rule.

Require literal dynamic-import paths

A computed import(expression) can bypass path resolution and therefore every subsystem boundary. Apply this rule to every production and test source block:

js
{
    files: ['**/*.{js,jsx,mjs,ts,tsx}'],
    rules: {
        'no-restricted-syntax': [
            'error',
            {
                selector: 'ImportExpression[source.type!="Literal"]',
                message:
                    'Dynamic import paths must be string literals so dependency boundaries remain enforceable.',
            },
        ],
    },
}

This permits import('./known-module.js') and rejects variables, concatenation, and template expressions. Keep the rule in the same flat config as the boundaries rules so a new source block cannot silently omit it.

Prevent production imports of test-only code

Declare test code as a narrower subsystem before the production catch-all, then forbid the production subsystem from importing it:

js
export default {
    subsystems: [
        { name: 'app-test', pattern: 'src/**/*.test.{js,jsx,ts,tsx}' },
        { name: 'app-test-support', pattern: 'test/**' },
        { name: 'app-prod', pattern: 'src/**' }, // catch-all, last
    ],
    forbidden: [
        {
            from: 'app-prod',
            to: 'app-test*',
            why: 'Production code must not import test-only code.',
        },
    ],
};

Tests may still import production subsystems. Adapt the globs to the repository, but retain the direction: production → test is forbidden.

Classify repository-root files

Root-level loose files — configs, entry scripts, generators — are the ones most often left unclassified, and the pattern's catch-all directive does not help here. A folder-mode ** at the root matches at the shallowest segment and claims files that specific elements already own, from any position in the list (see § 5). Declare the root's files with mode: 'file' instead:

js
// eslint.architecture.mjs (repository root)
export default {
    subsystems: [
        { name: 'repo-tooling-config', pattern: '*.config.js', mode: 'file' },
        { name: 'repo-tooling-entry',  pattern: 'server.js',   mode: 'file' },
        // No '**' catch-all at this level.
    ],
};

boundaries/no-unknown-files then names each root file still unclassified, so the list comes from the linter's output rather than from memory.

4. Output Contract

When applying this implementation, emit:

Scope:          <repo / package / subsystem path>
Decision:       Add eslint.architecture.mjs | Update assembler | Update ESLint config | Blocked
Generated config:<path, if any>
Rules changed:  <boundaries/dependencies element edges, externals package policy, or no-restricted-imports entries>
Verification:   <eslint command / assembler command / red-first proof / Not run + reason>
Next action:    <specific file edit, dependency install, or unresolved question>

5. JavaScript-specific gotchas

[!NOTE] Unresolved imports bypass enforcement. eslint-plugin-boundaries only enforces rules on imports it can resolve to a file path. Host-served absolute paths (e.g. SWA's /js/...) aren't resolved by default and pass silently. Fix: install eslint-import-resolver-alias and add it under settings['import/resolver'] so /js → packages/.../js resolves.

[!IMPORTANT] The rule block's files glob is a second, independent gate. A correct element registry, correct forbidden edges, and strict severities do nothing on a path outside files. A narrowed glob — ['packages/**/*.{js,ts,mjs}'], or an allowlist that grew one directory at a time — lets an entire new top-level directory into the repository with the boundaries lint reporting zero violations, because the rule never ran on it. Nothing in the output distinguishes that from a clean run. Keep the scope at ['**/*.{js,jsx,mjs,ts,tsx}'].

[!IMPORTANT] mode decides what a pattern matches, and the default is folder. In folder mode a pattern is tested against a file's path ancestors, so ** matches at the shallowest segment and captures files that deeper, more specific elements already own — regardless of its position in the list. A root-level { name: 'repo-unclassified', pattern: '**' } declared last, after several dozen specific elements, still wins, and yields a flood of misclassifications rather than the intended safety net. Use mode: 'file' for anything a shallower pattern would swallow.

[!NOTE] mode is deprecated in v7 and the suggested replacement does not cover this case. The deprecation warning recommends partialMatch: false; that does not classify a repository-root file — with a glob or with an exact filename, the file stays unknown. mode: 'file' is currently the only form that works. Accept the warning instead of chasing it.

[!NOTE] boundaries/external still works and is deprecated in v7. Unlike the mode deprecation above, this one has a working replacement: fold package policy into boundaries/dependencies with checkAllOrigins: true and the to.module.origin / to.module.source sub-selector (§3, Confine a provider SDK to its adapter). Do not run both rules over the same package — two verdicts on one edge, and the one that reports depends on rule order.

[!NOTE] Unmatched files bypass enforcement — silently. Files matching no element are invisible to boundaries/dependencies. End every constrained subsystem's subsystems list with a <dir>/** catch-all (pattern Directive 5), and set boundaries/no-unknown-files to error so an unmatched file is reported instead of ignored. Dependency rules are not a substitute: a file with no imports, or one loaded by a <script> tag, has no edge to judge.

[!NOTE] The plugin forbids; it never requires. An aspect coverage check — every subsystem in the governed set uses the mechanism — has no boundaries/dependencies form. dependency-cruiser's required rules can express it over the same registry; defect-shift-left places that check.

[!IMPORTANT] Flat config replaces a rule's options per rule id. Two blocks whose files overlap and that both set no-restricted-syntax or no-restricted-imports leave only the later block's entries active, and lint still passes. Merge the entries in the assembler.

[!NOTE] Emitted options are copied for every linted file. One selector per category can grow the boundaries options to megabytes and most of the lint time. Group element names per rule, measure uncached lint time after an assembler change, and guard the emitted size with a test.

[!NOTE] Facade-as-file pattern. JavaScript idiomatically exposes a facade as a single index/entry file. Use mode: 'file' plus an exact-path pattern (no glob) so the facade is matched alone.

[!NOTE] .js config breaks discovery. Many source-discovery walkers (the assembler's own, plus some lint plugins) treat .js as analyzable source. Naming the architecture file .js triggers self-reference and mis-classification. .mjs is required.

Bundled files

The model reads these on demand while the skill is loaded. They are exposed as readable files and are never executed.

Frequently asked questions

What does the Architecture As Code Javascript AI skill do?

JavaScript / TypeScript implementation of the `architecture-as-code` pattern. Per-subsystem `eslint.architecture.mjs` files merged into a single ESLint flat-config and enforced via `eslint-plugin-boundaries`. TRIGGER when: implementing or extending architecture-as-code in a JS/TS repo, debugging an `eslint-plugin-boundaries` rule, or adapting the assembler. SKIP for routine edits inside a governed subsystem. Reads in conjunction with `architecture-as-code` (the pattern, source of truth for schema, rule placement, anti-patterns, and audit checklist) — this skill defines only the JS-specific...

Why use Architecture As Code Javascript on TypingMind?

Because you install it once and use it with any model. Architecture As Code Javascript is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Architecture As Code Javascript in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/l-gevity/l-gevity-skills/tree/main/.agents/skills/architecture-as-code-javascript. TypingMind reads its SKILL.md and bundles its files and installs it as a skill you can enable per chat.

Which AI models can use Architecture As Code Javascript?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Architecture As Code Javascript?

As many as you like. As long as a model supports skills, you can use Architecture As Code Javascript with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Architecture As Code Javascript AI skill free?

Yes. It is published on GitHub by l-gevity under the MIT license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇