AGENT SKILL

aomi-transact

Build natural-language crypto/DeFi agents and EVM MCP plugins (Claude Code, Cursor, Codex, Gemini). Aomi turns prompts into wallet-signed txs on Ethereum, Base, Arbitrum, Optimism, Polygon, Linea — non-custodial, fork-simulated. 40+ apps: Uniswap, Aave, Lido, Morpho, GMX, Hyperliquid, Polymarket.

License 许可
MIT
所属仓库
ASoudy/AGS
最近更新
2026-05-23
Stars
★ 0
Forks
0

01 前言 Front Matter

name
aomi-transact
risk
critical
tags
[ "defi", "wallet", "account-abstraction", "cli", "eip-712", "onchain", "agent", "intent" ]
source
aomi-labs/skills (MIT)
license
MIT
date_added
2026-05-06
description
Build natural-language crypto/DeFi agents and EVM MCP plugins (Claude Code, Cursor, Codex, Gemini). Aomi turns prompts into wallet-signed txs on Ethereum, Base, Arbitrum, Optimism, Polygon, Linea — non-custodial, fork-simulated. 40+ apps: Uniswap, Aave, Lido, Morpho, GMX, Hyperliquid, Polymarket.
source_repo
aomi-labs/skills
license_source
https://github.com/aomi-labs/skills/blob/main/LICENSE

03 SKILL.md 正文

Aomi Transact

Authorized use only. This skill signs and broadcasts on-chain transactions on the user's behalf. The user must explicitly request each signing step. The skill will not run aomi tx sign without an explicit user request and a corresponding tx-N queued by aomi tx list.

Signing gate. Do not include aomi tx sign in a copied or runnable multi-command block. Stop after listing or simulating queued transactions, summarize the tx ids, chain, value, recipient, calldata purpose, and simulation result, then ask the user for an explicit signing instruction such as sign tx-1. Only run the exact signing command after that separate approval.

Overview

aomi-transact is a procedure for driving the Aomi CLI (@aomi-labs/client) from natural-language prompts. The user types something like "swap 1 ETH for USDC on Uniswap"; the agent picks the right protocol and contract, stages the approve+swap as a batch, simulates it on a forked chain, and returns a queued wallet request for the user to sign. The wallet only ever sees calldata that already passed simulation.

The CLI is account-abstraction-first: by default it signs through a zero-config Alchemy proxy (no provider credentials needed), using EIP-7702 on Ethereum mainnet and ERC-4337 on L2s. Each aomi <subcommand> invocation starts, runs, and exits — there is no long-running process.

The full skill including references (account-abstraction.md, apps.md, examples.md, session.md, troubleshooting.md, drain-vectors.md), templates (aomi-workflow.sh), and per-host metadata (agents/openai.yaml) lives upstream at aomi-labs/skills. This entry is the canonical SKILL.md only — clone the upstream for the full bundle.

When to Use This Skill

  • The user wants to chat with the Aomi agent from the terminal.
  • The user wants balances, prices, routes, quotes, or transaction status.
  • The user wants to build, simulate, confirm, sign, or broadcast wallet requests.
  • The user wants to simulate a batch of pending transactions before signing.
  • The user wants to inspect or switch apps, models, chains, or sessions.
  • The user wants to inspect or change Account Abstraction settings (EIP-7702 / ERC-4337).
  • The user wants to sign EIP-712 typed-data payloads (off-chain agreements, intent fillers).

Examples

Read-only — price check

aomi --prompt "what is the price of ETH?" --new-session

Returns a quote with no wallet request queued. Use aomi tx list to confirm there's nothing pending.

Single-tx flow — Lido stake

aomi chat "Stake 0.01 ETH with Lido to get stETH" \
  --public-key 0xUserAddress --chain 1 --new-session
aomi tx list

submit(address(0)) on Lido stETH 0xae7ab96520DE3A18E5e111B5EaAb095312D7fE84, value = 0.01 ETH. No approve, single tx. Stop here, show the queued transaction details, and wait for the user's explicit instruction before signing.

Multi-step batch — Uniswap V3 swap

aomi chat "swap 1 USDC for WETH on Uniswap V3, send to my wallet" \
  --public-key 0xUserAddress --chain 1 --new-session
aomi tx list                        # tx-1 = approve, tx-2 = swap
aomi tx simulate tx-1 tx-2          # mandatory for multi-step

The simulator runs each tx sequentially on a forked chain so the swap step sees the approve's state changes. Don't sign step 2 independently — it would revert. Stop after simulation, summarize the batch, and wait for an explicit user instruction naming both tx ids before signing.

Cross-chain — CCTP Ethereum → Base

aomi chat "Bridge 50 USDC from Ethereum to Base via CCTP. Recipient is my wallet." \
  --public-key 0xUserAddress --chain 1 --new-session
aomi tx list
aomi tx simulate tx-1 tx-2

Stop after simulation and wait for the user to explicitly approve signing the named tx ids. After signing, source-chain burn confirms in 1-2 blocks; destination mint requires Circle's off-chain attestation (~13-19 minutes).

Limitations

  • Requires @aomi-labs/client v0.1.30 or newer. Older versions lack --aa, --aa-provider, --aa-mode and the simulation gate. Install with npm install -g @aomi-labs/client or run on demand via npx @aomi-labs/client@0.1.30 ....
  • Active backend connection. The skill drives a CLI that talks to api.aomi.dev. Without network access, only local read commands (aomi tx list, aomi session log) work.
  • AA sponsorship on L2s is not guaranteed. The zero-config proxy path does not reliably sponsor on Base/Arbitrum/Optimism in v0.1.30. If the EOA has 0 native gas on the destination chain, aomi tx sign returns viem's insufficient funds for transfer. Either fund the EOA with a small amount of native gas, or configure a real BYOK Alchemy/Pimlico provider with a sponsorship policy. Do not retry with --eoa — that path also needs gas.
  • Per-session secret ingestion. Apps that require provider tokens (binance, polymarket, dune, etc.) must have credentials configured by the user in their own shell or via aomi secret add NAME=<value>. The skill never sets credentials on its own initiative.
  • Drain vectors are guard-blocked. The agent rejects calldata where recipient/onBehalfOf/mintRecipient ≠ msg.sender. This is a security feature, not a bug — surface the block to the user rather than reformulating the prompt.
  • Network/RPC failures. Public RPCs may rate-limit (429) or fail auth (401). The user must supply a reliable chain-matching RPC via --rpc-url for production signing.
  • Slippage and deadlines on live transactions. Quotes from deadline-bearing routes (Across, Khalani fillers) can expire while the user is reviewing; the agent self-heals by rebuilding with fresh deadlines, but the user should re-check aomi tx list for the latest passing batch.

Best Practices

  • Default --new-session on the first command of a new task. Reusing it mid-task starts a fresh conversation and the agent loses the quote it just gave you.
  • Always aomi tx list before aomi tx sign. Never assume a chat response queued a transaction.
  • Always aomi tx simulate tx-1 tx-2 ... before signing a multi-step batch. Single-tx flows are simulation-optional but never wrong to simulate.
  • Keep signing commands out of runnable examples. Show or run aomi tx sign only after the user gives a separate, explicit approval naming the exact queued tx-N ids.
  • Sign only Batch [...] passed txs. Skip orphans from earlier failed attempts (failed at step N: 0x...).
  • Match --rpc-url to the queued tx's chain, not the session chain (--chain) — they are independent controls.
  • Never echo credential values. The skill confirms credential setup with handle name or derived address only.

Authorization Disclaimer

This skill can sign and broadcast on-chain transactions worth real value. Use only on accounts you own and on networks you trust. The skill does not custody funds; the user retains full control of signing keys via --public-key and the underlying wallet. Review every queued tx-N before running aomi tx sign.

Source

Additional Resources

For the full skill including per-flow examples (CCTP bridge, Aave supply, Lido stake, Uniswap swap), AA mode reference, drain-vector table, troubleshooting guide, and the bash workflow template, see the upstream repo:

04 同仓库更多 Skills

ASoudy
ASoudy/AGS

web-typography

Select, pair, and implement typefaces for web projects. Use when the user mentions "font pairing", "which typeface", "line height", "responsive typography", "web font loading", "type hierarchy", "variable fonts", "FOUT/FOIT", or "typographic scale". Also trigger when choosing between system fonts and web fonts, optimizing font loading performance, or designing readable long-form content layouts. Covers readability evaluation, CSS implementation, and performance optimization. For overall UI design systems, see refactoring-ui. For dramatic typographic experiences, see top-design.

★ 0
ASoudy
ASoudy/AGS

ux-heuristics

Evaluate and improve interface usability using heuristic analysis. Use when the user mentions "usability audit", "UX review", "users are confused", "heuristic evaluation", "form usability", "navigation problems", "Nielsen heuristics", "cognitive walkthrough", or "usability testing". Also trigger when reviewing a design for usability issues, improving form completion rates, or evaluating information architecture and navigation. Covers Nielsens 10 heuristics, severity ratings, and information architecture. For visual design fixes, see refactoring-ui. For conversion-focused audits, see cro-methodology.

★ 0
ASoudy
ASoudy/AGS

traction-eos

Implement the Entrepreneurial Operating System (EOS) to align vision and execution across a company. Use when the user mentions "EOS", "V/TO", "quarterly rocks", "Level 10 meetings", "accountability chart", "IDS process", "Entrepreneurial Operating System", or "business operating system". Also trigger when a growing company needs meeting structure, goal-setting frameworks, or a systematic approach to solving recurring organizational issues. Covers the six EOS components: Vision, People, Data, Issues, Process, Traction. For team motivation design, see drive-motivation. For lean experimentation, see lean-startup.

★ 0
ASoudy
ASoudy/AGS

top-design

Create award-winning, immersive web experiences at the level of Awwwards-featured agencies. Use when the user mentions "premium website", "portfolio site", "scroll animations", "Awwwards quality", "brand experience", "cinematic web design", "parallax storytelling", or "agency-quality site". Also trigger when building landing pages that need to impress, designing creative portfolios, or elevating a standard website to a memorable digital experience. Covers dramatic typography, purposeful motion, scroll-based composition, and performance-optimized animation. For foundational UI, see refactoring-ui. For type selection, see web-typography.

★ 0
ASoudy
ASoudy/AGS

system-design

Design scalable distributed systems using structured approaches for load balancing, caching, database scaling, and message queues. Use when the user mentions "system design", "scale this", "high availability", "rate limiter", "design a URL shortener", "system design interview", "capacity planning", or "distributed architecture". Also trigger when estimating infrastructure requirements, choosing between microservices and monoliths, or designing for millions of concurrent users. Covers common system designs and back-of-the-envelope estimation. For data fundamentals, see ddia-systems. For resilience, see release-it.

★ 0
ASoudy
ASoudy/AGS

storybrand-messaging

Clarify brand messaging using narrative structure that positions the customer as hero. Use when the user mentions "brand message", "website copy", "elevator pitch", "one-liner", "messaging isnt resonating", "brand script", "StoryBrand framework", or "customer as hero". Also trigger when rewriting homepage copy, crafting email nurture sequences, or creating consistent messaging across sales collateral and marketing materials. Covers landing page copy, marketing collateral, and consistent communication. For memorable messaging, see made-to-stick. For product positioning, see obviously-awesome.

★ 0