meyverickmeyverick/agentic/skills/framework-engine-stacks/SKILL.md

framework-engine-stacks

Skill0 starsChanged 13 days agoISCView on GitHub

Recommend and enforce the preferred three-tier architecture, engine pairings, and domain stacks for full-stack, simulation, and native targets. Use when scaffolding new modules, choosing technical stacks, evaluating frontend or backend frameworks, or configuring Rust-Axum-SvelteKit integration.

In this repository18 other files it gives its agents.
- AGENTS.mdAGENTS.md
- apex-performance-auditorSkill
- checkSkill
- context7-mcpSkill
- design-craftSkill
- guardrailsSkill
- hypermedia-engine-stacksSkill
- learning-curatorSkill
- okf-docsSkill
- pi-lens-ast-grepSkill
- pi-lens-lsp-navigationSkill
- pi-lens-write-ast-grep-ruleSkill
- pi-lens-write-tree-sitter-ruleSkill
- rust-systemsSkill
- scalabilitySkill
- select-modelSkill
- skill-authoring-visionSkill
- systems-auditorSkill

The file

We split this skill into the rules it gives an agent

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

≈0
Common · estimate
Many files like this have a rule of this kind.
≈2
Distinctive · estimate
This author's own choices. Worth a look.
3
Very specific
Tied to their project's names, files or tools. Probably skip.

Common
Distinctive
Very specific

Very specific:
Commit updated query metadata using cargo sqlx prepare to guarantee hermetic CI verification.
line 41
Very specific:
SSR Monoliths: Avoid Node.js-dependent SSR engines.
line 44
Distinctive:
Request-Blocking Compute: Never run heavy calculations or Candle inference inside request handlers.
line 45
Distinctive:
Client Polling: Do not poll APIs for dynamic updates.
line 46
Very specific:
Filesystem Coupling: Avoid cross-tier relative directory traversal (../) at runtime.
line 47

≈ “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

Framework & Engine Stacks

Summary

Guidance and selection rules for scaffolding the workspace's preferred three-tier stack across web, backend compute, and graphics/simulation workloads. Every tier prioritizes native performance, static runtime delivery, and end-to-end type safety.

Core Mental Model & Preferred Tiers

UI & Web Tier: SvelteKit static SPA via @sveltejs/adapter-static (fallback: 'index.html') styled with Tailwind v4 and Svelte 5 Runes, served directly by Axum tower-http.
State Coordinator & Storage: Axum and Tokio in Rust managing database schema via SQLx migrations in project/api/migrations/, backed by PostgreSQL. Shared contracts export via ts-rs directly to project/web/src/lib/types/bindings/.
Compute & In-Process ML: Pure Rust utilizing Tokio work-stealing, Rayon worker pools for heavy background execution, and Candle for zero-Python local GGUF/Safetensors inference.
Client Simulation & Graphics: Miniplex client-side ECS for dynamic entity lifecycles paired with modular rendering adapters: Threlte for declarative 3D, Babylon.js for WebGPU/Havok physics, PixiJS for high-density 2D (>1,000 nodes), or Phaser for turnkey arcade systems.
Cross-Platform Shell: Tauri v2 packaging the compiled static SPA for desktop and mobile targets.

Architectural Selection Matrix

Workload Domain	Recommended Architecture & Engine	Rationale & Trade-offs
Full-Stack Web SPA	SvelteKit + Tailwind v4 + Axum `tower-http`	Eliminates SSR node fragmentation by serving static SPA routes from the Axum binary.
Relational Persistence	PostgreSQL + SQLx (`project/api/migrations/`)	Compile-time query verification via `sqlx-data.json` and strict expand-contract migrations.
High-Density 2D (>1,000 nodes)	Miniplex ECS + PixiJS	Zero-allocation frame queries paired with batched WebGL 2D rendering.
Complete 2D Arcade Engine	Miniplex ECS + Phaser	Turnkey physics, tilemaps, and audio managers decoupled from domain logic.
Declarative 3D Scenes	Miniplex ECS + Threlte	Reactive Svelte 5 component trees integrated with Three.js viewports.
Advanced 3D & WebGPU Physics	Miniplex ECS + Babylon.js	Native WebGPU shaders and Havok rigid-body simulation.
Multi-Core Background Compute	Rust (Axum + Rayon)	Dedicated worker pools preventing async Tokio reactor blocking.
In-Process ML & Vector Search	Rust + Candle	Embedded model execution without Python runtimes or external dependencies.
Desktop & Mobile Targets	Tauri v2	Minimalist native webview wrapping the pre-built static export.

Operational Workflow

Scope Intake & Contract Definition
- Identify whether the deliverable is a web client, spatial simulation, compute job, or desktop shell.
- Define backend domain events and derive TS types via ts-rs into project/web/src/lib/types/bindings/ before authoring views.
Mounting & Fallback Configuration
- Configure @sveltejs/adapter-static with fallback: 'index.html'.
- Mount Axum API, WebSocket, and gRPC endpoints strictly before attaching the tower-http static fallback handler.
Local Tooling & CI Verification
- Proxy /api and /ws in vite dev to cargo watch -x run during active iteration.
- Commit updated query metadata using cargo sqlx prepare to guarantee hermetic CI verification.

Anti-Patterns & Edge Cases

SSR Monoliths: Avoid Node.js-dependent SSR engines; deploy purely static SPA distributions served from Axum.
Request-Blocking Compute: Never run heavy calculations or Candle inference inside request handlers; route jobs to Rayon workers.
Client Polling: Do not poll APIs for dynamic updates; use PostgreSQL LISTEN/NOTIFY and Axum WebSockets/SSE channels.
Filesystem Coupling: Avoid cross-tier relative directory traversal (../) at runtime; share data solely via typed network protocols.

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