Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Use when a user needs a QR code or wallet payment URI to fund a Superior-managed EVM wallet on a specific chain before using Lighter, Polymarket, Hyperliquid, or other Superior Trade workflows.
.claude/skills/superior-trade-deposit-qr/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-07 | ✗→✓ | ▲ Improved | -49% | 0% |
| case-09 | ✗→✓ | ▲ Improved | -47% | 0% |
| case-13 | ✗→✓ | ▲ Improved | 1% | 0% |
| case-17 | ✗→✓ | ▲ Improved | 41% | 0% |
| case-20 | ✗→✓ | ▲ Improved | 171% | 0% |
Use this skill to create a wallet-scannable QR code for sending a coin or ERC-20 token to a Superior-managed EVM wallet address on a named chain.
This is for funding the wallet itself. It is not the same as a venue-specific deposit intent. For Lighter, a user may first need USDC in a payer wallet, but the Lighter account still requires POST /v3/portfolio/lighter/deposit to get the Lighter CCTP intent address.
For external bridge/provider flows such as Relay, Robinhood Chain USDG, MetaMask Mobile bridge links, or prefilled third-party deposit URLs, use skills/external-deposit instead. Do not use this primitive to encode arbitrary bridge calldata, approvals, or contract deposit calls into transaction QRs.
Ask for or verify all four fields:
0x1111111111111111111111111111111111111111arbitrum, base, avalanche, polygon, or ethereumUSDC, USDC.e, ETH, AVAX, MATIC, or a token contract addressNever guess the chain. The same address can exist on many EVM chains, and sending a token on the wrong chain can strand funds.
transfer payment URIs.Use the bundled script:
bashnode skills/deposit-qr/scripts/create-deposit-qr.mjs \ --address 0x1111111111111111111111111111111111111111 \ --chain arbitrum \ --asset USDC \ --amount 5
It prints JSON with:
payment_uri — EIP-681 URI for the requested token transferqr_url — hosted QR image URL for the payment URIplain_address_qr_url — fallback QR image URL containing only the destination addresssummary — human-readable chain, asset, token, amount, and destinationDisplay the QR in Markdown when the client supports images:
markdown
bashnode skills/deposit-qr/scripts/create-deposit-qr.mjs \ --address 0x1111111111111111111111111111111111111111 \ --chain arbitrum \ --asset USDC \ --amount 5
Tell the user:
textDeposit QR Summary: * Destination: 0x1111111111111111111111111111111111111111 * Chain: Arbitrum One * Asset: native USDC * Token contract: 0xaf88d065e77c8cC2239327C5EDb3A432268e5831 * Amount: 5 USDC Scan the QR with a wallet that supports Arbitrum token-transfer URIs. If the wallet does not prefill the token transfer correctly, use the plain address fallback and manually select Arbitrum native USDC.
skills/external-deposit for prefilled bridge links and quote-backed external wallet flows.After the wallet has the correct USDC, use the lighter skill to create the Lighter CCTP deposit intent:
textPOST /v3/portfolio/lighter/deposit
That endpoint returns the actual Lighter intent address to receive the USDC transfer for venue credit.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-07 | fail→pass | 16,885 | 3,105 | -82% | 1 | 1 | 0% | 3,703 | 1,903 | -49% | 0 | 0 | — |
case-01 | fail→fail | 7,233 | 7,469 | +3% | 1 | 1 | 0% | 1,533 | 1,658 | +8% | 0 | 0 | — |
case-02 | fail→fail | 6,625 | 6,247 | -6% | 1 | 1 | 0% | 1,370 | 1,621 | +18% | 0 | 0 | — |
case-03 | fail→fail | 8,307 | 19,753 | +138% | 1 | 1 | 0% | 1,684 | 5,593 | +232% | 0 | 0 | — |
case-04 | pass→fail | 9,171 | 5,563 | -39% | 1 | 1 | 0% | 2,132 | 1,586 | -26% | 0 | 0 | — |
case-05 | pass→pass | 6,516 | 7,463 | +15% | 1 | 1 | 0% | 1,175 | 2,620 | +123% | 0 | 0 | — |
case-06 | fail→fail | 12,770 | 8,972 | -30% | 1 | 1 | 0% | 2,369 | 1,640 | -31% | 0 | 0 | — |
case-08 | pass→pass | 10,700 | 5,621 | -47% | 1 | 1 | 0% | 1,814 | 2,216 | +22% | 0 | 0 | — |
case-09 | fail→pass | 26,758 | 6,705 | -75% | 1 | 1 | 0% | 4,444 | 2,335 | -47% | 0 | 0 | — |
case-10 | pass→fail | 7,878 | 7,268 | -8% | 1 | 1 | 0% | 1,648 | 1,902 | +15% | 0 | 0 | — |
case-11 | pass→fail | 9,638 | 6,213 | -36% | 1 | 1 | 0% | 2,117 | 1,642 | -22% | 0 | 0 | — |
case-12 | pass→pass | 8,687 | 3,400 | -61% | 1 | 1 | 0% | 1,432 | 1,664 | +16% | 0 | 0 | — |
case-13 | fail→pass | 12,385 | 8,543 | -31% | 1 | 1 | 0% | 2,541 | 2,570 | +1% | 0 | 0 | — |
case-14 | fail→fail | 17,246 | 7,541 | -56% | 1 | 1 | 0% | 3,721 | 1,633 | -56% | 0 | 0 | — |
case-15 | pass→pass | 8,919 | 3,668 | -59% | 1 | 1 | 0% | 1,450 | 1,814 | +25% | 0 | 0 | — |
case-16 | fail→fail | 7,724 | 7,581 | -2% | 1 | 1 | 0% | 1,688 | 1,628 | -4% | 0 | 0 | — |
case-17 | fail→pass | 7,574 | 2,925 | -61% | 1 | 1 | 0% | 1,115 | 1,569 | +41% | 0 | 0 | — |
case-18 | fail→fail | 12,803 | 8,702 | -32% | 1 | 1 | 0% | 2,294 | 1,753 | -24% | 0 | 0 | — |
case-19 | pass→pass | 9,930 | 20,553 | +107% | 1 | 1 | 0% | 1,607 | 4,997 | +211% | 0 | 0 | — |
case-20 | fail→pass | 6,141 | 7,586 | +24% | 1 | 1 | 0% | 905 | 2,456 | +171% | 0 | 0 | — |
case-21 | fail→pass | 8,629 | 5,442 | -37% | 1 | 1 | 0% | 1,411 | 2,087 | +48% | 0 | 0 | — |
case-22 | fail→pass | 18,093 | 5,270 | -71% | 1 | 1 | 0% | 2,846 | 2,134 | -25% | 0 | 0 | — |
case-23 | fail→pass | 14,380 | 3,091 | -79% | 1 | 1 | 0% | 2,589 | 1,741 | -33% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 23 cases were attempted, and 14 counted toward the lift figure. The other 9 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +22 percentage points is the difference between those two pass rates over the 14 comparable cases. 3 cases got worse with the skill loaded, and they are included in that figure.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.