LLMS - Ethan Daya

# Ethan Daya

> Founder and software engineer building at the edge of software, agents, and the real world. Based in New York City.

## Agent Skills

Reusable instructions for agents to work like Ethan and meet his expectations across design, engineering, collaboration, and model-assisted review. They are also public artifacts for understanding how he approaches the work.

Install with the Skills CLI:

```bash
npx skills add https://ethandaya.com --skill collaborating-with-ethandaya engineering-with-ethandaya designing-with-ethandaya
```

- [designing with ethan daya](https://ethandaya.com/skills/designing-with-ethandaya/SKILL.md): reference-driven restraint, strong hierarchy, and verified polish
- [engineering with ethan daya](https://ethandaya.com/skills/engineering-with-ethandaya/SKILL.md): root-cause correctness, narrow scope, and honest evidence
- [collaborating with ethan daya](https://ethandaya.com/skills/collaborating-with-ethandaya/SKILL.md): concise communication, decisive judgment, and understated personality
- [consulting fable oracle](https://ethandaya.com/skills/consulting-fable-oracle/SKILL.md): Amp-only adversarial reviewer, devil's advocate, and repository research partner for Amp's native Oracle

## Agent Setup

Reviewed July 2026. Repository guidance and the current request govern, followed by Ethan's first-party skills, then optional third-party skills. Keep installs project-scoped, target only the active agent, and never use wildcards with third-party sources.

### foundation

Ethan's first-party operating guidance. Install for every workspace.

- [collaborating-with-ethandaya](https://ethandaya.com/skills/collaborating-with-ethandaya/SKILL.md): Outcome-first communication, decisions, critique, and handoffs.
- [engineering-with-ethandaya](https://ethandaya.com/skills/engineering-with-ethandaya/SKILL.md): Root-cause implementation, narrow scope, and honest evidence.
- [designing-with-ethandaya](https://ethandaya.com/skills/designing-with-ethandaya/SKILL.md): Reference-locked direction, responsive continuity, and rendered verification.

### recommended companions

Portable browser verification and skill discovery when installation is allowed.

- [agent-browser](https://skills.sh/vercel-labs/agent-browser/agent-browser): Exercise real browser behavior and capture runtime evidence in fresh sandboxes.
- [find-skills](https://skills.sh/vercel-labs/skills/find-skills): Discover specialized skills only when the work requires them instead of installing a broad catalog.

```bash
npx skills add vercel-labs/agent-browser --agent <agent-id> --skill agent-browser -y
npx skills add vercel-labs/skills --agent <agent-id> --skill find-skills -y
```

### frontend craft

Install for polished interface work. Use emil-design-eng for general component craft and motion decisions, including timing and easing, then prefer the specialist guidance: finishing details from make-interfaces-feel-better, color systems from oklch-skill, named transition recipes from transitions-dev, and WCAG 2.2 conformance from accessibility. Use WCAG ratios—not APCA—for formal contrast claims, and prefer component-scoped reduced-motion behavior over a global reset.

- [emil-design-eng](https://skills.sh/emilkowalski/skill/emil-design-eng): Apply broad design-engineering judgment, improve component feel, and guide motion timing and easing.
- [make-interfaces-feel-better](https://skills.sh/jakubkrehel/make-interfaces-feel-better/make-interfaces-feel-better): Apply Jakub Krehel's focused guidance for typography, surfaces, optical alignment, and interaction details.
- [oklch-skill](https://skills.sh/jakubkrehel/oklch-skill/oklch-skill): Build perceptually consistent palettes, migrate color tokens, handle gamut, and repair contrast with OKLCH.
- [transitions-dev](https://skills.sh/jakubantalik/transitions.dev/transitions-dev): Install production-ready CSS transition recipes with semantic tokens, implementation hooks, and reduced-motion guards.
- [accessibility](https://skills.sh/addyosmani/web-quality-skills/accessibility): Apply WCAG 2.2 semantics, keyboard, focus, forms, contrast, assistive-technology, and testing guidance.

```bash
npx skills add emilkowalski/skill --agent <agent-id> --skill emil-design-eng -y
npx skills add jakubkrehel/make-interfaces-feel-better --agent <agent-id> --skill make-interfaces-feel-better -y
npx skills add jakubkrehel/oklch-skill --agent <agent-id> --skill oklch-skill -y
npx skills add jakubantalik/transitions.dev --agent <agent-id> --skill transitions-dev -y
npx skills add addyosmani/web-quality-skills --agent <agent-id> --skill accessibility -y
```

### vercel pack

Install for React and Next.js implementation, component architecture, performance, and interface review.

- [vercel-react-best-practices](https://skills.sh/vercel-labs/agent-skills/vercel-react-best-practices): Apply current React and Next.js performance patterns during implementation and review.
- [vercel-composition-patterns](https://skills.sh/vercel-labs/agent-skills/vercel-composition-patterns): Design scalable React component APIs without boolean-prop proliferation or unnecessary abstractions.
- [web-design-guidelines](https://skills.sh/vercel-labs/agent-skills/web-design-guidelines): Audit interface code for usability, accessibility, and web design quality.

```bash
npx skills add vercel-labs/agent-skills --agent <agent-id> --skill vercel-react-best-practices vercel-composition-patterns web-design-guidelines -y
```

### cloudflare pack

Install for Workers and Cloudflare platform work; add narrower product skills on demand.

- [cloudflare](https://skills.sh/cloudflare/skills/cloudflare): Route Cloudflare work through current product documentation and platform guidance.
- [workers-best-practices](https://skills.sh/cloudflare/skills/workers-best-practices): Author and review Workers code against production runtime practices.
- [wrangler](https://skills.sh/cloudflare/skills/wrangler): Use current Wrangler commands and deployment conventions.

```bash
npx skills add cloudflare/skills --agent <agent-id> --skill cloudflare workers-best-practices wrangler -y
```

### railway pack

Install for Railway deployments, services, databases, environments, domains, and platform operations.

- [use-railway](https://skills.sh/railwayapp/railway-skills/use-railway): Operate Railway infrastructure through current CLI, MCP, deployment, configuration, and verification guidance.

```bash
npx skills add railwayapp/railway-skills --agent <agent-id> --skill use-railway -y
```

### plugin tooling

Install project-scoped Amp plugins only when their prerequisites are already available. Inspect any existing file at the target path before replacing it, then confirm each plugin appears in `amp plugins list` before relying on its tools.

- [Fable Oracle](https://ethandaya.com/plugins/fable-oracle.ts): Challenge Amp Oracle conclusions, find counterevidence, and research overlooked alternatives in the repository with a read-only Claude Fable 5 partner.
  - Prerequisite: Amp plus Claude Code 2.1.207 or newer, authenticated with access to Claude Fable 5.

```bash
mkdir -p .amp/plugins && curl -fsSL https://ethandaya.com/plugins/fable-oracle.ts -o .amp/plugins/fable-oracle.ts
```

Amp's native Oracle remains the primary oracle. Install and load [consulting-fable-oracle](https://ethandaya.com/skills/consulting-fable-oracle/SKILL.md) in Amp only to challenge its assumptions, seek falsifying evidence, and research overlooked alternatives in the repository—not to compete with it or replace it. Give Fable a differentiated adversarial or research assignment instead of sending both tools the same prompt. Fable runs the local Claude Code CLI, so usage follows that authentication and subscription rather than consuming Amp inference credits, but that cost difference must not determine which review the task requires. Do not install the skill directly into Claude Code, Codex, or another harness.

### harnesses

Use Amp first. Its native Oracle remains the primary oracle. Pair it with Claude Code through Fable Oracle as an adversarial reviewer, devil's advocate, and repository research partner—not as competition or a substitute. Use Codex directly when it is the active harness; no Codex plugin is required.

- Amp + Claude Code via Fable Oracle plugin: Claude Fable 5 with high reasoning effort
  - Amp only skill: [consulting-fable-oracle](https://ethandaya.com/skills/consulting-fable-oracle/SKILL.md)

```bash
npx skills add https://ethandaya.com --agent amp --skill consulting-fable-oracle -y
```

- Codex via native harness: GPT-5.5 with xhigh reasoning effort

Use these harness and model pairings only when selectable without changing user or global settings. Otherwise keep the current harness and use the strongest available reasoning model. Do not interrupt or restart work solely to switch harnesses or models.

## Currently

- CTO and co-founder, [Scene Infrastructure Company](https://sceneinfrastructure.com) ([mesh.ing](https://mesh.ing))

## Previously

- Consultant and advisor across fintech, product, and infrastructure
- Co-creator, [Zora](https://zora.co)
- Product developer, [Upgrowth](https://upgrowth.com.au)
- Freelance product designer and developer across telecom and advertising

## Projects

- [Revue](https://revue.run)
- [Tourminal](https://tourminal.cc)

## Writing

- the context window is closing (draft)

## Featured

- [A16z-backed scene infrastructure company hopes to show crypto is “not just money”](https://blockworks.com/news/a16z-backed-social-dao), Blockworks, 2024
- [Haun Ventures leads $50M round in NFT startup Zora Labs](https://techcrunch.com/2022/05/05/haun-ventures-leads-50m-round-in-nft-startup-zora-labs/), TechCrunch, 2022
- [Zora raises $2M to build a sustainable creator economy](https://techcrunch.com/2020/10/16/crypto-driven-marketplace-zora-raises-2m-to-build-a-sustainable-creator-economy/), TechCrunch, 2020

## Online

- [X](https://x.com/ethandaya)
- [GitHub](https://github.com/ethandaya)
- [LinkedIn](https://www.linkedin.com/in/ethan-daya-743817142/)
- [Email](mailto:me@ethandaya.com)
- [RSS](https://ethandaya.com/rss.xml)

## Canonical

- Human-readable portfolio: https://ethandaya.com
- Machine-view documentation: https://ethandaya.com/llms
- Machine-readable profile: https://ethandaya.com/llms.txt