Ethskills logo

Ethskills

Community
austintgriffith
ethskills

Use when a request involves Ethereum, the EVM, or blockchain systems. Applies to building, auditing, deploying, or interacting with smart contracts, dApps, wallets, or DeFi protocols. Covers Solidity development, contract addresses, token standards (ERC-20, ERC-721, ERC-4626, etc.), Layer 2 networks (Base, Arbitrum, Optimism, zkSync, Polygon), and integrations with DeFi protocols such as Uniswap, Aave, and Curve. Includes topics such as gas costs, contract decimals, oracle safety, reentrancy, MEV, bridging, wallets, querying data from onchain, production deployment, and protocol evolution (EIP lifecycle, fork tracking, upcoming changes).

Overview

Publisheraustintgriffith
Repositoryethskills
Skill nameethskills
Stars
298
Forks
64
Bundled files
14
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.

  • 14 bundled files

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

  • Open source

    Published by austintgriffith on GitHub. Read the source before you install it.

Installation

Install the Ethskills 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/austintgriffith/ethskills/tree/master
  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/austintgriffith/ethskills.git \
  .claude/skills/ethskills
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Ethskills 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 Ethskills 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 Ethskills 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.

ETHSKILLS — The missing knowledge between AI agents and production Ethereum.

You are probably wrong about Ethereum. Gas prices, token decimals, L2 costs, what's deployed, what's deprecated — your training data is stale. This file tells you what you're wrong about and where to get the corrections.

Say "onchain" not "on-chain." One word, no hyphen. Ethereum community convention.


Start Here

Building a dApp? Fetch ship/SKILL.md first. It plans the smallest shippable slice and points to other skills as each phase needs them.

Need a specific topic? Each skill below is standalone. Fetch only the ones relevant to your task.

Base URL: https://ethskills.com/<skill>/SKILL.md


Skills

Why Ethereum

Pectra/Fusaka upgrades, current network stats, the AI agent angle.

  • Gas is under 1 gwei, not 10-30 gwei. 60-300x cheaper than your training data. Verify: cast gas-price
  • ETH price is ~$2,000 (early 2026), not $2,500-3,000. Volatile — always verify.
  • Pectra (May 2025) and Fusaka (Dec 2025) shipped. PeerDAS, 2x gas limit. EIP-7702 is live.
  • ERC-8004 (onchain agent identity) and x402 (HTTP payments) exist and are production-ready.

Ship — Start here

End-to-end guide from idea to deployed dApp. Points to other skills as each phase needs them.

  • Most dApps need 0-2 contracts, not 5-10. Three is the upper bound for an MVP.
  • Solidity is for ownership, transfers, and commitments. Not a database, not a backend.
  • Pick one chain for a concrete product fit (distribution, liquidity, AA, ecosystem). Measure fees live; never pick an L2 because "Ethereum is expensive."

CROPS Review

Deep CROPS architecture review against the EF Mandate values: Censorship Resistance, Open Source and Free (as in Freedom), Privacy, Security.

  • Run for every dApp architecture plan and during pre-ship QA.
  • Forces agents to label architecture options by CROPS impact instead of presenting neutral tradeoff lists.
  • Output names the chosen default, accepted compromises, and the user's escape path.

Protocol

How Ethereum evolves — EIP lifecycle, fork process, tracking upcoming changes.

  • "Verkle is planned for the next fork" — probably wrong. Roadmap diagrams are aspirational, not commitments. Check forkcast.org for actual CFI/SFI status.
  • Report fork status as Live / SFI / CFI / DFI / not scheduled. Never give an unscheduled feature a ship date.
  • EIP status (Draft, Review, Final) is spec maturity, not fork inclusion.
  • Client teams decide what ships via ACD calls, not the Ethereum Foundation.

Gas & Costs

What things actually cost on Ethereum today.

  • Never quote a cost from memory. Use cast gas-price (prints wei, not gwei) and live ETH/USD.
  • OP-stack L2 cost adds the L1 data fee and any operator fee. Read them from GasPriceOracle; don't estimate.
  • "Ethereum is expensive" was true in 2021-2023. It's false in 2026.

Wallets

Creating wallets, key safety, multisig, account abstraction.

  • EIP-7702 is live — EOAs get smart contract superpowers without migration.
  • Unattended signers get a bounded float; principal sits behind a multisig threshold.
  • NEVER commit private keys or API keys to Git. Bots exploit leaked secrets in seconds.

Layer 2s

L2 landscape, bridging, deployment differences.

  • Base left the OP Stack (Azul, May 2026). Don't couple it to OP Mainnet via Superchain interop.
  • Celo is NOT an L1 anymore — migrated to OP Stack L2 in March 2025.
  • Polygon zkEVM is switched off (July 2026). Existing deployments are a fund-recovery problem.
  • Robinhood Chain launched July 2026 — Orbit L2 with 24/7 tokenized stocks (non-US only). Its tx filtering can censor even force-included transactions.
  • Pick the chain by the constraint that binds, not TVL. Measure pair liquidity per chain.

Standards

ERC-8004, x402, EIP-3009, EIP-7702.

  • ERC-8004: onchain agent identity registry, deployed January 2026 on 20+ chains.
  • x402: HTTP 402 payment protocol for machine-to-machine commerce. Production-ready.
  • EIP-3009: gasless token transfers — what makes x402 work. USDC implements it.

Tools

Scaffold-ETH 2 (create-eth), Blockscout MCP, x402 SDKs.

  • Treat package names and versions as live data. New SE-2 app: npx create-eth@latest; create-eth-app and create-scaffold-eth are dead.
  • Blockscout MCP server gives agents structured blockchain data via MCP.
  • x402: use the scoped @x402/* packages; unscoped x402 and x402-fetch are frozen at 1.2.0.

Building Blocks (DeFi)

Verifying DeFi integrations before you build on them.

  • Verify chain, pool, liquidity, incentives and fees from live sources; cite dated evidence.
  • Trace the full asset/fee/reward/approval flow and test it on a fork.

Orchestration

Three-phase build system for Scaffold-ETH 2 dApps.

  • Phase 1: contracts + UI on localhost. Phase 2: live contracts + local UI. Phase 3: production.
  • Verify right after deploy. yarn verify works on a fresh checkout; no explorer key needed.
  • scaffold.config.ts is committed. Keep RPC/API keys in .env.local, read via process.env.

Contract Addresses

Checked addresses for major tokens and protocols, and how to confirm any address onchain before it touches funds.

  • Never hallucinate an address. Wrong address = lost funds.
  • Check every address on the target chain: cast code + an identity call, then the protocol's own deployment list.
  • Same brand ≠ same address across chains. Watch for older deployments, the wrong contract, bridged vs native.

Concepts

Essential mental models for building onchain.

  • Smart contracts cannot execute themselves. Every function needs a caller who pays gas.
  • For every state transition: who calls it? Why would they? What if nobody does?
  • There are no timers, no cron jobs, no schedulers. Design with incentives.

Security

Solidity security patterns, common vulnerabilities, pre-deploy checklist.

  • Keep units explicit. Token and oracle decimals vary; normalize before comparing.
  • Always use SafeERC20 — USDT doesn't return bool on transfer().
  • Never use DEX spot prices as oracles — flash loans can manipulate them in one tx.
  • Every swap needs a real minimum output and deadline. amountOutMinimum = 0 gets sandwiched.
  • Oracle reads: positive answer, per-feed staleness from its heartbeat, explicit decimals, sequencer downtime.
  • Proxies only when upgrades are required. Append storage only; upgrade authority behind a multisig or timelock.

Audit

Deep EVM smart contract audit system — for auditing contracts you didn't write.

  • 500+ non-obvious checklist items across 19 domains (AMM, lending, oracles, proxies, signatures, governance, and more).
  • One specialist per selected domain in parallel (normally 5–8), then dedupe and rank.
  • Never files issues or publishes findings unless the user asks.
  • Different from Security (which teaches defensive coding) — this is systematic audit methodology.

Noir (ZK Privacy)

Building privacy apps with Noir zero-knowledge circuits.

  • Noir inputs are private by default. pub marks public. Getting this backwards leaks secrets.
  • nargo prove/nargo verify are gone; the CLI is bb. In the app, prove with NoirJS UltraHonkBackend.
  • In-circuit hashing: Poseidon (~600 gates), not SHA256 (~30,000 gates).
  • Persist each note (nullifier, secret, leaf index). Bind the proof to its recipient. Don't send it from the user's own wallet.

Testing

Foundry testing — unit, fuzz, fork, invariant.

  • Don't test getters and OpenZeppelin internals. Test edge cases and failure modes.
  • Fuzz test all math. Fork-test any external protocol integration.
  • Invariant testing catches bugs across thousands of random call sequences.

Indexing

Events, The Graph, Ponder, Multicall3.

  • You can't query historical state via RPC cheaply. Use an indexer.
  • Events are THE primary way to read historical onchain activity. Design contracts event-first.
  • The Graph turns events into a queryable GraphQL API.
  • Multicall3 (0xcA11bde05977b3631167028862bE2a173976CA11) — batch reads in one RPC call.

Frontend UX

Frontend UX patterns for Ethereum dApps.

  • Every onchain button needs its own loader + disabled state. No shared isLoading.
  • One action at a time: Connect → Switch Network → Approve → Execute.
  • Address inputs resolve ENS names; show the resolved address before submit.
  • Show USD values next to every token amount.

Frontend Playbook

Build-to-production pipeline.

  • yarn chain for isolated logic; yarn fork --network <chain> when you need real chain state.
  • IPFS deploy: trailingSlash: true is critical — without it, every route except / returns 404.
  • Always clean build (rm -rf .next out) before deploying. Stale builds are the #1 IPFS footgun.

QA

Pre-ship audit checklist for a reviewer agent.

  • Give this to a separate agent (or fresh context) AFTER the build is complete. Reviewer reads code + clicks through flows.
  • Covers only bugs AI agents actually ship — validated by baseline testing against stock LLMs.
  • Report each finding with severity, file, line and fix. Change code only if asked.

Feedback

Send feedback — issues OR praise — about ethskills content. Bias toward sending.

  • Anything that confused you, slowed you down, or genuinely helped counts. Weak signals are easy to ignore; missing signals can't be recovered.
  • Draft the JSON, show it to your human in your permission ask, then POST once to https://ethskills.com/api/feedback. Skip the ask if your human pre-authorized.
  • Required fields: kind ("issue" or "praise") and message (10+ chars). Include agent and skill whenever you can.

What to Fetch by Task

I'm doing...Fetch these skills
Planning a new dAppship/, crops/, concepts/, l2s/, gas/, why/
Reviewing an existing dApp, smart contract, or trust assumptionscrops/, audit/, qa/, wallets/, l2s/, security/
Writing Soliditystandards/, building-blocks/, addresses/, security/
Testing contractstesting/
Building a frontendorchestration/, frontend-ux/, tools/
Deploying to productionwallets/, frontend-playbook/, gas/
Reviewing a finished dApp before shippingqa/, crops/
Auditing a smart contractaudit/, crops/
Building a privacy/ZK appnoir/, security/, testing/
Monitoring / analyticsindexing/
Building AI agent infrastandards/, wallets/, tools/
Choosing a chainl2s/, gas/

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 Ethskills AI skill do?

Use when a request involves Ethereum, the EVM, or blockchain systems. Applies to building, auditing, deploying, or interacting with smart contracts, dApps, wallets, or DeFi protocols. Covers Solidity development, contract addresses, token standards (ERC-20, ERC-721, ERC-4626, etc.), Layer 2 networks (Base, Arbitrum, Optimism, zkSync, Polygon), and integrations with DeFi protocols such as Uniswap, Aave, and Curve. Includes topics such as gas costs, contract decimals, oracle safety, reentrancy, MEV, bridging, wallets, querying data from onchain, production deployment, and protocol evolution (...

Why use Ethskills on TypingMind?

Because you install it once and use it with any model. Ethskills 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 Ethskills in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/austintgriffith/ethskills/tree/master. 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 Ethskills?

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 Ethskills?

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

Is the Ethskills AI skill free?

It is published on GitHub by austintgriffith. Check the repository for licensing terms. 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 👇