agentleFS
kaynetikkaynetik/skills/ultimate-web3/SKILL.md

ultimate-web3

Skill1 starsChanged 8 months agoMITView on GitHub

Architecture and protocol-design layer for production Web3. Covers chain selection (EVM L1/L2, Solana, Cosmos, TON), Solidity architecture (proxies, diamonds, factories), L2/rollup stack choice (OP Stack, Arbitrum Orbit, ZK stacks, EIP-4844), account abstraction (ERC-4337, EIP-7702, ERC-6900), bridges (LayerZero, CCIP, Wormhole, intents), oracles and indexers (Chainlink, Pyth, Ponder, Envio), MEV and order-flow, custody (multisig, MPC, Safe), privacy L1 (zk-SNARKs, stealth addresses), governance (Governor, timelocks), deployment ops (CREATE2, Tenderly, Defender), and architectural anti-patterns. Use when designing a protocol, choosing a chain or rollup, planning cross-chain or multi-chain systems, hardening custody, planning deployment and incident response, designing governance, or when the user mentions architecture, proxy, diamond, OP Stack, Arbitrum Orbit, ERC-4337, EIP-7702, LayerZero, CCIP, Chainlink, MEV, multisig, MPC, Safe, Tenderly, Defender, or "which chain should I use".

The file

We split this skill into the rules it gives an agent

10 rules. About 0 are common, 6 are distinctive, and 4 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈6

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 4

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    If the question is about how to build something at the implementation layer, route to the specialist skill above.

    line 23
  2. Distinctive:

    Do not paste back code the user already shared.

    line 27
  3. Distinctive:

    Read at most 2 reference files per turn.

    line 28
  4. Distinctive:

    Never inline content from a companion skill (frontend code, test code, vulnerability classes).

    line 29
  5. Distinctive:

    Prefer tables, decision trees, and matrices over prose.

    line 30
  6. Very specific:

    Use WebSearch for L2 rankings, TVL, bridge security incidents, and active EIPs.

    line 31
  7. Very specific:

    When the user asks "which chain / which rollup / which bridge / which oracle", read reference/chain-selection.md first.

    line 32
  8. Very specific:

    Verify before recommending: Use WebSearch for current L2 stage classifications, TVL, bridge incidents, EntryPoint version, and active EIPs.

    line 60
  9. Very specific:

    If you only have time to read one tree, read this.

    line 86
  10. Distinctive:

    If the question fits one of the above, route there and stop reading this skill.

    line 117

≈ “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

Ultimate Web3 (Architecture & Strategy)

Architect-level skill for cross-chain, cross-discipline Web3 design decisions. The body is intentionally thin; depth lives in reference/. Open only the files you actually need.

Companion skills (route to these first)

This skill is the architecture layer. Defer to companion skills for their domains:

DomainUse skill
dApp frontend (wagmi, viem, RainbowKit, SIWE, Next.js dApps)web3-frontend
Smart contract testing (Foundry, Hardhat, fuzz, invariants, fork)web3-testing
Solidity vulnerability classes and secure coding patternssolidity-security
Codebase security audit, pen test, or findings reportsecurity-audit
Whitepaper / litepaper authoringwhitepapers
Public-facing token / staking / governance copy reviewlegalizer

If the question is about how to build something at the implementation layer, route to the specialist skill above. Use this skill for what to build, on which platform, with which architecture, and how to operate it.

Token economy rules for the agent

  1. Do not paste back code the user already shared.
  2. Read at most 2 reference files per turn. Choose by the routing table below.
  3. Never inline content from a companion skill (frontend code, test code, vulnerability classes); link to the companion skill instead.
  4. Prefer tables, decision trees, and matrices over prose.
  5. Use WebSearch for L2 rankings, TVL, bridge security incidents, and active EIPs. Do not memorize dynamic facts.
  6. When the user asks "which chain / which rollup / which bridge / which oracle", read reference/chain-selection.md first; do not improvise a recommendation.

Routing table

User intent or keywordOpen file
"which chain", "EVM vs Solana", "L1 vs L2", "app-chain", platform tradeoffsreference/chain-selection.md
Proxy, UUPS, transparent proxy, diamond, ERC-2535, factory, modular contracts, hooksreference/solidity-architecture.md
OP Stack, Arbitrum Orbit, zkSync, Linea, Scroll, Base, app-chain vs general L2, EIP-4844, blobreference/l2-and-rollups.md
Solana, Anchor, PDA, CPI, Cosmos, CosmWasm, IBC, app-chain SDK, TON, Tact, FunCreference/non-evm.md
ERC-4337, EntryPoint, bundler, paymaster, EIP-7702, ERC-6900, session keys, gasless, smart accountreference/account-abstraction.md
Bridge, LayerZero, Wormhole, CCIP, Axelar, Across, intent-based, lock-and-mint, burn-and-mintreference/bridges-and-interop.md
Oracle, Chainlink, Pyth, Redstone, UMA, TWAP, indexer, The Graph, Ponder, Envio, Goldskyreference/oracles-indexers.md
MEV, sandwich, frontrun, private mempool, Flashbots, MEV-Boost, builder, OFA, CoW, UniswapX, 1inch Fusionreference/mev-and-flow.md
Custody, multisig, Safe, Gnosis, MPC, Fireblocks, HSM, threshold signatures, deploy keys, hot walletreference/custody-and-keys.md
Privacy L1, zk-SNARK, zk-STARK, ring signature, stealth address, Bulletproof, Tornado, Aztec, Aleo, encrypted statereference/privacy-l1.md
Governance, Governor, Snapshot, Tally, timelock, delegate, voting power, conviction voting, futarchyreference/governance.md
Deployment, CREATE2, deterministic deploy, Etherscan verification, monitoring, Tenderly, Defender, Forta, incident responsereference/deployment-ops.md
"is this an anti-pattern", "footgun", architecture review, common mistakesreference/anti-patterns.md

Golden rules (non-negotiable)

  1. Custody is design, not configuration. Decide hot/warm/cold separation, multisig/MPC, deploy-key isolation, and rotation policy before writing the first line of contract code. See reference/custody-and-keys.md.
  2. Choose the rollup stack before the chain. OP Stack vs Arbitrum Orbit vs ZK stacks affects bridges, gas economics, blob inclusion, and decentralization roadmap more than the brand name. See reference/l2-and-rollups.md.
  3. Bridges are your weakest link. Cross-chain messaging has caused most of the largest exploits in Web3 history. If you can avoid bridging, do. If you must bridge, prefer canonical bridges or intent-based systems with explicit settlement guarantees. See reference/bridges-and-interop.md.
  4. Oracles are part of your trust model. Push vs pull, single oracle vs aggregate, manipulation cost, staleness checks, and TWAP windows are protocol-level decisions, not integration details. See reference/oracles-indexers.md.
  5. Upgradeability is a liability. Proxies are not free: storage layout, initializer races, admin compromise, and selector clashes are all real. Pin a clear upgrade strategy (none / UUPS / diamond / migrate-by-deploy) and document it. See reference/solidity-architecture.md.
  6. Deployment is a protocol moment, not a sysadmin task. Multi-chain CREATE2, deterministic addresses, verification across explorers, post-deploy monitoring, and pause/freeze paths must exist on day one. See reference/deployment-ops.md.
  7. Verify before recommending. Use WebSearch for current L2 stage classifications, TVL, bridge incidents, EntryPoint version, and active EIPs. Ecosystem facts change month-to-month.

Default stack matrix

Defaults to use unless the user has different constraints. Document any deviation.

LayerDefaultWhen to deviate
Smart contract language (general)Solidity, latest stable (>= 0.8.26)Vyper for math-heavy DeFi, Yul for hot paths
EVM contract frameworkFoundryTypeScript-first team -> Hardhat 3 (route to web3-testing)
Solidity standard libraryOpenZeppelin Contracts (audited release)Solady for gas-critical primitives
Upgradeability (when needed)UUPS via OpenZeppelin UpgradesDiamond (ERC-2535) only when modular plugin extensibility is a real requirement
MultisigSafe (formerly Gnosis Safe)Custom only with a strong reason and an audit
Governance contractsOpenZeppelin Governor + TimelockSnapshot-only off-chain when no on-chain execution
Oracle (price)Chainlink Data FeedsPyth for high-frequency / non-EVM, Redstone for low-cap assets, UMA for optimistic
Cross-chain messagingNative L1<->L2 canonical bridgesLayerZero / CCIP only with explicit security analysis
IndexerPonder or Envio (TS, self-hosted)The Graph for decentralized indexing requirements
MonitoringOpenZeppelin Defender + Tenderly Web3 ActionsForta detection bots for high-value protocols
L2 (consumer app, EVM)BaseArbitrum for DeFi composability, Optimism for OP Stack ecosystem
L2 (DeFi, EVM)Arbitrum OneBase for Coinbase distribution
Non-EVM high-throughputSolana (Anchor)Aptos / Sui only with explicit ecosystem reason
App-chain (cosmos ecosystem)Cosmos SDK + CosmWasmPolkadot parachain only with cross-parachain requirement
Wallet for end usersSmart account (EIP-7702 enhanced EOA where supported, ERC-4337 otherwise)Plain EOA only for power users / dev tooling

Chain decision shortcut

If you only have time to read one tree, read this. For depth, see reference/chain-selection.md.

Constraint -> Default chain
|
|-- "EVM compatibility required"
|   |-- "Coinbase / consumer distribution"     -> Base
|   |-- "DeFi composability, deepest TVL"      -> Arbitrum One
|   |-- "ZK finality required"                 -> zkSync Era / Linea / Scroll / StarkNet
|   |-- "Need own chain, EVM"                  -> OP Stack rollup or Arbitrum Orbit
|   `-- "L1 only, conservative"                -> Ethereum mainnet
|
|-- "Highest throughput, lowest cost"
|   |-- "Rust team"                            -> Solana (Anchor)
|   `-- "Move team"                            -> Aptos or Sui (verify ecosystem fit first)
|
|-- "Sovereign chain, IBC interop"             -> Cosmos SDK + CosmWasm
|-- "Telegram-native distribution"             -> TON (Tact)
`-- "Privacy-first, novel L1"                  -> Custom design (read privacy-l1.md)

When NOT to apply this skill

  • Frontend implementation: web3-frontend (wagmi, viem, RainbowKit, SIWE, transaction UX).
  • Test writing: web3-testing (Foundry, Hardhat, fuzz, invariants, fork).
  • Solidity vulnerability prevention at the function level: solidity-security.
  • Repository vulnerability audit or findings report: security-audit.
  • Whitepaper structure and citations: whitepapers.
  • Public-facing token / governance copy review: legalizer.
  • Pure cryptography research (new ZK schemes, novel signature constructions): out of scope.

If the question fits one of the above, route there and stop reading this skill.

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.