agentleFS
aomi-labsaomi-labs/skills/aomi-transact/SKILL.md

aomi-transact

Skill8 starsChanged 6 months agoMITView on GitHub

Build natural-language crypto agents, web3 assistants, and trading bots that read and write EVM chain state. Aomi turns prompts ("swap 1 ETH for USDC", "open a 3x GMX long", "bet $100 on Polymarket") into wallet-signed transactions on Ethereum, Base, Arbitrum, Optimism, Polygon, Linea — non-custodial, fork-simulated. Use when the user wants a crypto/DeFi agent, AI trading/wallet assistant, or on-chain execution against Uniswap, Aave, Lido, GMX, Hyperliquid, Polymarket, Binance, OKX, or 40+ other protocols. Trigger with prompts about swaps, lending, bridging, staking, perps, prediction markets, or any DeFi/CEX action needing a wallet signature. Supports durable Commit review and backend-prepared EVM or AA signing. MUST NOT fabricate or echo credentials; values reach the CLI only when the user explicitly supplied them.

The file

We split this skill into the rules it gives an agent

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

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈3

    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
  1. Very specific:

    Use the TypeScript aomi CLI to prepare, review and execute user-authorized transactions.

    line 48
  2. Distinctive:

    Establish capabilities: Read commands.md first. Check package availability and installed help.

    line 53
  3. Very specific:

    Execute the reviewed Commit: Use aomi tx sign <commit-id> only within the user's authorization and an available designated signer.

    line 77
  4. Distinctive:

    Use the existing authorized signer without exposing key material.

    line 93
  5. Very specific:

    Use https://aomi.chat as the production origin.

    line 115
  6. Distinctive:

    Use a fresh client session to discover both tool sets, then perform a read-only session-list or catalog call.

    line 130

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 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

Aomi Transact

Use the TypeScript aomi CLI to prepare, review and execute user-authorized transactions. The Rust aomi-cli in product-mono has a different interface.

Procedure

  1. Establish capabilities. Read commands.md first. Check package availability and installed help; a source test or SDK migration message does not prove a working published CLI. Stop a missing capability branch with its release requirement; keep a working older binary.
  2. Resolve funding. Identify the requested recipient, amount, destination chain ID and token contract. Read the user's wallet catalog for operational addresses and known funded chains. Check live token and native gas balances across those positions before preparing a payment. A zero Ethereum USDC balance is not a wallet-wide shortfall. Use tx preflight when available; follow funding checks otherwise. Keep a treasury/Safe refill as a separately authorized source choice.
  3. Bind the intent. Show the selected sender, source asset/chain, destination asset/chain, amount, recipient, estimated fees and shortfall. For conversion or bridging, obtain a fresh quote with minimum destination output, gas on each leg and route fees. USDC and USDG are different contracts/assets. Finish this step with unambiguous, funded terms before building a payment.
  4. Prepare once. On a new task use aomi chat "<exact intent>" --new-session --public-key <sender> --chain <chain-id>. Continue its original session on follow-ups. Read aomi tx list --json, retain the full Commit ID and export that exact Commit with aomi tx export <commit-id> --format commit.
  5. Review. Match the durable review to the approved intent, including sender, chain, token, amount, recipient, calldata, fees and simulation. Check all legs and approval limits. Simulation for a Commit lives in its review; tx simulate is a separate legacy Action tool. Require passed simulation for execution.
  6. Execute the reviewed Commit. Use aomi tx sign <commit-id> only within the user's authorization and an available designated signer. For an external signer, preserve the exact export and report the signed payloads or prepared broadcast hash through tx submit; see workflows.md. Inspect prepared AA payloads before choosing the signing path; see account-abstraction.md.
  7. Reconcile the same payment. Retain original session, full Commit ID, approved intent and transaction hash together. Refresh that Commit with aomi tx list --json before retrying; the backend observes the receipt and moves it to confirmed. A known hash or outstanding wallet attempt calls for reconciliation, not a new signing attempt. Report preparation, simulation, signature, submission and confirmation separately. Bridge completion also needs the destination receipt and spendable output token.

Credentials and state

Use the existing authorized signer without exposing key material. Credential setup is a separate user-directed operation; values stay opaque and out of chat, logs, fixtures and exported payment records. ~/.aomi may contain signing keys, bearers and secret handles. Its permission manifest expresses intended scope; actual enforcement depends on the host. For credential handling, local state and recovery, read gotchas.md and session.md.

Branch references

  • 404, missing package, timeout or retry: troubleshooting.md.
  • Commands, installation and flags: commands.md.
  • Funding, execution and secret ingestion: workflows.md.
  • Direct payment or bridge walkthrough: examples.md.
  • Protocol field/recipient warnings: drain-vectors.md.
  • App discovery and prompt examples: apps.md; runtime capabilities and this procedure take precedence over historical app examples.

Source: https://github.com/aomi-labs/skills/tree/main/aomi-transact

Hosted MCP

Use https://aomi.chat as the production origin. If your CLI still targets the retired host, run aomi config set-backend https://aomi.chat and update any AOMI_BACKEND_URL or --backend-url override. For an existing MCP connection, change its URL and authorize again; old-origin OAuth grants do not carry over.

For hosted tools, connect Agent and Pipeline separately with OAuth:

codex mcp add aomi-agent --url https://aomi.chat/v1/agent/mcp
codex mcp add aomi-pipeline --url https://aomi.chat/v1/pipeline/mcp

Codex derives the resource from each URL; omit --oauth-resource to avoid a second resource parameter. If the authorization page reports a CIMD metadata fetch error, retry that server's login with --oauth-client-registration dcr. Use a fresh client session to discover both tool sets, then perform a read-only session-list or catalog call. Enabled configuration alone is not a successful smoke.

Each grant must have the exact connected resource URL as its audience. Agent, Pipeline, REST and guest credentials are not interchangeable. OAuth grants tool access; wallet execution still follows the approved Commit and designated signer. A browser GET can return 404 because these resources accept POST and OPTIONS.

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.