Lifi logo

Lifi

OrganizationPopular
internet-court
lifi

LI.FI REST API for cross-chain and same-chain token swaps, bridging, DeFi deposits (Composer), yield discovery (Earn), and intent-based execution (Intents). USE THIS SKILL WHEN USER WANTS TO: - Swap tokens between different blockchains (e.g., "swap USDC on Ethereum to ETH on Arbitrum") - Bridge tokens to another chain (e.g., "move my ETH from mainnet to Optimism") - Swap tokens on the same chain with best rates (e.g., "swap ETH to USDC on Polygon") - Find the best route or quote for a token swap across chains - Deposit into DeFi vaults/lending/staking in one click, including cross-chain (Composer: Aave, Morpho, Pendle, EtherFi, Yearn, etc.) - Discover yield opportunities, vault APY/TVL data, or track DeFi positions (Earn) - Execute gasless or intent-based transfers via a solver network (LI.FI Intents) - Move stablecoins cheaply with optimized defaults (stablecoin preset) - Build multi-chain payment flows (accept any token, settle in specific token) - Check supported chains, tokens, bridges, or gas prices - Track status of a cross-chain transaction, recover from failed/partial transfers - Query transfer history/analytics or withdraw collected integrator fees - Build backends, bots, or AI agents (any language) that need cross-chain functionality

Overview

Publisherinternet-court
Repositoryinternet-court-skill
Skill namelifi
Stars
5.8K
Forks
106
Bundled files
1
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.

  • 1 bundled files

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

  • Open source

    Published by internet-court on GitHub. Read the source before you install it.

Installation

Install the Lifi 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.
  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/internet-court/internet-court-skill.git /tmp/internet-court-skill
mkdir -p .claude/skills
cp -r /tmp/internet-court-skill/vendored/lifi/lifi .claude/skills/lifi
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Lifi 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 Lifi 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 Lifi 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.

LI.FI API Integration

LI.FI aggregates bridges, DEXs, intent solvers, and DeFi protocols behind one API. This skill covers the full product surface:

ProductBase URLWhat it does
Core API (swaps & bridges)https://li.quest/v1Quotes, routes, execution data, status tracking across 75+ chains
Composer (DeFi execution)https://li.quest/v1 (same /quote endpoint)One-click swap/bridge + deposit/withdraw into vaults, lending, staking
Earn (yield data)https://earn.li.fi/v1Vault discovery, APY/TVL analytics, portfolio positions
Intents (solver marketplace)https://order.li.fiIntent-based orders filled by a competitive solver network

For AI agents, LI.FI also offers an MCP server (https://mcp.li.quest/mcp) and a CLI (@lifi/cli). LI.FI recommends agents and backends use the REST API directly rather than the SDK.

Authentication & Rate Limits

API key is optional — it only raises rate limits. Header: x-lifi-api-key (get one at the LI.FI Partner Portal).

bash
curl "https://li.quest/v1/chains" -H "x-lifi-api-key: YOUR_API_KEY"
Tier/quote, /advanced/routes/advanced/stepTransactionOther endpoints
Without API key75 req / 2 hours50 req / 2 hours100 req / minute
With API key (default)100 RPM (2-hour rolling window: 12,000 / 2h)same100 RPM

Responses include ratelimit-limit, ratelimit-remaining, ratelimit-reset (seconds) headers. 429 responses carry error code 1005. Never expose your API key in client-side code.

Quick Start — The Five-Call Recipe

The canonical flow for any swap/bridge:

1. GET /chains   → discover chains (id, key, chainType, nativeToken)
2. GET /tokens   → find tokens & decimals (?chains=1,42161)
3. GET /quote    → get quote with ready-to-sign transactionRequest
4. [Execute]     → if ERC-20: check allowance vs estimate.approvalAddress, approve if needed;
                   then sign & send transactionRequest
5. GET /status   → poll until DONE or FAILED
bash
# 3. Quote (fromToken/toToken accept addresses OR symbols)
curl "https://li.quest/v1/quote?\
fromChain=42161&toChain=10&\
fromToken=0xaf88d065e77c8cC2239327C5EDb3A432268e5831&\
toToken=0xDA10009cBd5D07dd0CeCc66161FC93D7c9000da1&\
fromAmount=10000000&fromAddress=0xYourAddress&slippage=0.005"

# 5. Status (pass fromChain to speed it up; bridge = quote's `tool`)
curl "https://li.quest/v1/status?txHash=0xYourTxHash&fromChain=42161&toChain=10&bridge=across"

From the quote response, extract: transactionRequest (sign & send), estimate.approvalAddress (ERC-20 spender), estimate.toAmount/toAmountMin, estimate.executionDuration (basis for polling), and tool (pass as bridge to /status). Quotes go stale in ~60 seconds — re-fetch before signing if older (and always after a separate approval tx).

Core Endpoints

GET /quote

Single best route with transaction data ready for execution.

ParameterTypeRequiredDescription
fromChain / toChainstringYesChain ID or key (same value = same-chain swap)
fromToken / toTokenstringYesToken address or symbol
fromAmountstringYesAmount in smallest unit
fromAddressstringYesSender wallet address
toAddressstringNoRecipient (defaults to fromAddress)
slippagenumberNo0.005 = 0.5% (default)
orderstringNoFASTEST or CHEAPEST
integratorstringNoYour app ID (analytics + fee collection)
feenumberNoIntegrator fee (0.02 = 2%)
presetstringNoRouting preset, e.g. stablecoin (see below)
allowBridges / denyBridges / preferBridgesstring[]NoTool keys from /tools, or all/none/default
allowExchanges / denyExchanges / preferExchangesstring[]NoSame
allowDestinationCallbooleanNoDefault true
maxPriceImpactnumberNoDefault 0.10 (10%)
fromAmountForGasstringNoAmount converted to gas on destination
skipSimulationbooleanNoFaster response, less accurate gas limit
svmPriorityFeeLevelstringNoSolana priority fee: NORMAL/FAST/ULTRA
swapStepTimingStrategies / routeTimingStrategiesstring[]Noe.g. minWaitTime-600-4-300

Variants:

  • GET /quote/toAmount — pass toAmount instead of fromAmount; API computes the required input.
  • POST /quote/contractCalls — bridge + arbitrary destination-chain contract calls (manual calldata). For supported DeFi protocols, prefer Composer instead.

POST /advanced/routes + POST /advanced/stepTransaction

Multiple route options for comparison. Note the different naming: body uses fromChainId, fromTokenAddress (addresses only), with filters and preset inside options{}. Routes contain steps without transaction data — POST each step to /advanced/stepTransaction to populate transactionRequest.

Use /quote for simple transfers (1 call); use /advanced/routes when the user needs choices or price comparison (2+ calls).

GET /status

txHash (required — sending hash, receiving hash, or step id), plus optional fromChain (recommended, speeds up response), toChain, bridge.

Statuses: NOT_FOUNDPENDINGDONE | FAILED. On DONE, check substatus: COMPLETED, PARTIAL (different token received — full value preserved), or REFUNDED. See Status Recovery below.

Discovery & utility

  • GET /chains?chainTypes=EVM,SVM,UTXO,MVM,TVM. If omitted, returns EVM only. Non-EVM: Solana (SVM), Bitcoin (UTXO), Sui (MVM), Tron (TVM).
  • GET /tokens?chains=1,137&tags=stablecoin&minPriceUSD=0.01
  • GET /token?chain=POL&token=DAI
  • GET /tools — valid bridge/exchange keys for the allow*/deny* filters
  • GET /connections — available token-pair connections (at least one filter required)
  • GET /gas/prices, /gas/prices/{chainId}, /gas/suggestion/{chain}` — gas data; GET /gas/status — LIFuel tx status
  • GET /calldata/parse (beta) — ?callData=0x...&chainId=1 decodes LI.FI calldata into named function calls
  • Analytics: GET /v1/analytics/transfers (filtered, max 1000), GET /v2/analytics/transfers (cursor-paginated), GET /v1/analytics/transfers/summary (cross-chain amounts received, 30-day max range)
  • Integrator fees: GET /integrators/{integratorId} (collected fee balances), GET /integrators/{integratorId}/withdraw/{chainId} (withdrawal transactionRequest)

Deprecated — do not use: /advanced/possibilities, /gas/refetch, /analytics/wallets/{address}.

Routing Presets

preset=stablecoin (on /quote, or options.preset on /advanced/routes) applies optimized defaults for stablecoin transfers: order=CHEAPEST, slippage=0.001, price impact capped at 2%, and a shortlist of stablecoin-friendly bridges (Glacis, Mayan, Celer, Eco, Relay, Across, Gaszip). Explicit request params always override preset defaults — but list overrides like denyBridges replace the preset's lists entirely. Find qualifying tokens with GET /tokens?tags=stablecoin. (commerce preset coming soon.)

Composer — One-Click DeFi Deposits & Withdrawals

Composer bundles swap + bridge + protocol interaction (deposit, stake, lend, withdraw) into a single atomic transaction. There is no dedicated endpoint — set toToken to a supported protocol's vault/staking/deposit token address on the standard GET /quote, and Composer activates automatically:

bash
# 100 USDC on Arbitrum → Morpho Spark USDC vault on Base, one signature
curl "https://li.quest/v1/quote?\
fromChain=42161&toChain=8453&\
fromToken=0xaf88d065e77c8cC2239327C5EDb3A432268e5831&\
toToken=0x7BfA7C4f149E7415b73bdeDfe609237e29CBF34A&\
fromAmount=100000000&fromAddress=0xYou&toAddress=0xYou"
  • Detection: same-chain deposit quotes show the zap tool as top-level tool (e.g. composer, but the engine may route via other integrated zap tools such as fly); cross-chain quotes show the bridge as tool with the zap step in includedSteps. Don't hard-code on tool === "composer" — any quote with a vault address as toToken is a deposit route.
  • Withdrawals: reverse it — vault token as fromToken, desired token as toToken.
  • Every quote is simulated on a fork before being returned; failures surface as API errors instead of reverting txs.
  • Coverage: ~29 protocols (Aave V3, Morpho, Pendle, Spark, Euler, Fluid, Yearn, EtherFi, Kelp, Ethena, Maple, …) across 23 EVM chains. Some are deposit-only (Ethena, Kinetiq, Maple, Royco, USDai, Concrete, Ember) — withdraw via the protocol's own UI.
  • Limitations: EVM only; tokenized positions only (the target must mint a token); cross-chain flows are eventually consistent, not atomic — a destination deposit can fail after a successful bridge, leaving the user with the bridged token (poll /status, handle PARTIAL).
  • Use higher slippage (e.g. 0.01) for cross-chain composer routes.

See references/REFERENCE.md for the full protocol list and error codes.

Earn — Yield Discovery & Portfolio Data

Earn is a read-only data API at https://earn.li.fi (separate host!) normalizing vaults from 20+ protocols. Unlike the core API, it requires the x-lifi-api-key header (401 without). Execution is delegated to Composer: take a vault's address and use it as toToken on li.quest/v1/quote.

EndpointPurpose
GET /v1/vaultsList vaults — filter by chainId, asset (symbol or address), protocol, minTvlUsd, isTransactional, isRedeemable, isComposerSupported; sort with sortBy=apy|tvl; cursor pagination (limit ≤ 100, follow nextCursor)
GET /v1/vaults/{chainId}/{address}Single vault detail
GET /v1/chains / GET /v1/protocolsFilter dropdown data (Earn-scoped, not platform-wide)
GET /v1/portfolio/{userAddress}/positionsUser's DeFi positions with USD balances
bash
# Top USDC vaults on Base by APY, depositable via Composer
curl "https://earn.li.fi/v1/vaults?chainId=8453&asset=USDC&sortBy=apy&isComposerSupported=true&limit=10" \
  -H "x-lifi-api-key: YOUR_API_KEY"

Gotchas: boolean filters are string literals ("true"/"false"); APY values are decimals (0.0534 = 5.34%); TVL/balances are strings; lpTokens is usually empty — use the vault's top-level address for execution; gate deposit/withdraw UI on isTransactional/isRedeemable; check timeLock, caps, kyc fields. Rate limit: 50 req/min per key. Vault data refreshes every 15 min (transactional flags every 2 min).

Discover → deposit recipe: GET earn.li.fi/v1/vaults → pick vault → GET li.quest/v1/quote with vault address as toToken → approve + send → poll /status.

LI.FI Intents — Solver Marketplace

A separate intent-based system at https://order.li.fi (dev: order-dev.li.fi). Users express desired outcomes; solvers pre-fund delivery on the destination chain against standing quotes. Foundation of the Open Intents Framework. Integrator endpoints need no API key.

EndpointPurpose
POST /quote/requestQuote an intent (swapType: exact-input/exact-output); returns quotes[] best-first with quoteId
POST /orders/submitOff-chain order submission (Compact/gasless flows)
GET /orders/statusBy onChainOrderId or catalystOrderId
GET /ordersList/filter orders
GET /chains/supported / GET /routesLive coverage (EVM + Solana + Tron)

Two funding models:

  • Escrow (recommended default): approve → on-chain open(order) on InputSettlerEscrow — no /orders/submit call needed, solvers detect the Open event.
  • Compact (gasless): deposit once into The Compact, then sign EIP-712 BatchCompact orders off-chain and submit via /orders/submit.

Order lifecycle: Submitted/Open → Signed → Delivered → Settled (terminal). Refundable after expires if unfilled. Addresses use EIP-7930 interoperable format (chain ID embedded), not bare 0x addresses. Amounts in smallest units. See REFERENCE.md for StandardOrder structs, contract addresses, and signing details.

Note: the classic li.quest/v1/quote already aggregates intent-based bridges (Across, Relay, etc.) — only integrate order.li.fi directly when you want the order-server model itself (solver selection, resource locks, gasless submission).

Token Approvals

  • Native token (0x0000…0000) → no approval needed.
  • Spender is always quote.estimate.approvalAddressnever hardcode it (it varies by route).
  • Check allowance(fromAddress, approvalAddress); if < fromAmount, send approve and wait for confirmation before the main tx, then re-fetch the quote (gas estimates go stale).
  • USDT quirk: if current allowance > 0, you must approve(spender, 0) first, then approve the new amount.
  • Prefer exact-amount approvals for agents; approval gas ≈ 50–100k.

Status Tracking & Recovery

Poll GET /status with backoff: 10s (first ~6 attempts) → 30s → 60s → 120s. Use estimate.executionDuration from the quote to set expectations. Same-chain swaps are atomic and usually instant.

ResultAction
NOT_FOUND / PENDINGKeep polling (NOT_FOUND just means not indexed yet)
DONE + COMPLETEDSuccess
DONE + PARTIALUser received a different token (full value preserved, typically the bridge token on the destination). Optionally recover: quote a same-chain swap from receiving.token to the intended token; skip if gas > ~10% of value
DONE + REFUNDEDTokens returned on the source chain
FAILEDPermanent — check substatus (SLIPPAGE_EXCEEDED, OUT_OF_GAS, NOT_PROCESSABLE_REFUND_NEEDED, …), inform user

On timeout, never assume failure — save txHash and share lifiExplorerLink (https://explorer.li.fi/tx/<txHash>). Funds are non-custodial and end with the user; prevent PARTIAL by raising slippage for volatile tokens or targeting the bridge's native token directly.

Error Handling

HTTPRetry?Action
400NoFix parameters
404 / NO_POSSIBLE_ROUTEMaybeDifferent pair/amount/chains; try USDC/USDT/ETH as intermediary
429 (code 1005)YesExponential backoff min(2^attempt × 1s, 30s)
500/503Yes (1–2×)Wait 2–10s, retry

Common API error codes: 1002 NoQuote, 1007 Slippage (raise slippage to min(current×2, 0.03) or lower amount), 1011 Validation. Tool-level codes inside errors[]: NO_POSSIBLE_ROUTE, AMOUNT_TOO_LOW (Composer minimums run higher), INSUFFICIENT_LIQUIDITY, TOOL_TIMEOUT (retry immediately), FEES_HIGHER_THAN_AMOUNT. On execution reverted at send time: the quote was stale — fetch a fresh one (max ~2 retries). Full table in REFERENCE.md.

Integration Examples

Python

python
import requests

BASE_URL = "https://li.quest/v1"
HEADERS = {"x-lifi-api-key": "YOUR_API_KEY"}  # optional

def get_quote(from_chain, to_chain, from_token, to_token, amount, address, **kwargs):
    params = {
        "fromChain": from_chain, "toChain": to_chain,
        "fromToken": from_token, "toToken": to_token,
        "fromAmount": amount, "fromAddress": address,
        "slippage": 0.005, "integrator": "your-app", **kwargs,
    }
    r = requests.get(f"{BASE_URL}/quote", params=params, headers=HEADERS)
    r.raise_for_status()
    return r.json()

def get_status(tx_hash, from_chain=None, to_chain=None, bridge=None):
    params = {"txHash": tx_hash, "fromChain": from_chain,
              "toChain": to_chain, "bridge": bridge}
    return requests.get(f"{BASE_URL}/status", params=params, headers=HEADERS).json()

# Cross-chain swap quote
quote = get_quote(42161, 10,
    "0xaf88d065e77c8cC2239327C5EDb3A432268e5831",   # USDC.e Arbitrum
    "0xDA10009cBd5D07dd0CeCc66161FC93D7c9000da1",   # DAI Optimism
    "10000000", "0xYourAddress")

# Composer deposit: same call, vault address as toToken
deposit = get_quote(8453, 8453,
    "0x833589fCD6eDb6E08f4c7C32D4f71b54bdA02913",   # USDC Base
    "0x7BfA7C4f149E7415b73bdeDfe609237e29CBF34A",   # Morpho vault
    "100000000", "0xYourAddress")

# transactionRequest values are hex strings — parse before use
tx = quote["transactionRequest"]
gas_limit = int(tx["gasLimit"], 16)

Node.js

javascript
const BASE_URL = 'https://li.quest/v1';

async function getQuote(params) {
  const res = await fetch(`${BASE_URL}/quote?${new URLSearchParams(params)}`);
  if (!res.ok) throw new Error(`Quote failed: ${res.status} ${await res.text()}`);
  return res.json();
}

async function pollStatus(txHash, fromChain, toChain, bridge) {
  const params = new URLSearchParams({ txHash, fromChain, toChain, bridge });
  for (let attempt = 0; attempt < 60; attempt++) {
    const data = await (await fetch(`${BASE_URL}/status?${params}`)).json();
    if (data.status === 'DONE' || data.status === 'FAILED') return data;
    const delay = attempt < 6 ? 10_000 : attempt < 12 ? 30_000 : 60_000;
    await new Promise((r) => setTimeout(r, delay));
  }
  throw new Error('Status polling timed out — check explorer.li.fi');
}

// Earn: discover top vaults, then deposit via Composer
const { data: vaults } = await (await fetch(
  'https://earn.li.fi/v1/vaults?chainId=8453&asset=USDC&sortBy=apy&isComposerSupported=true&limit=5'
)).json();
const quote = await getQuote({
  fromChain: 8453, toChain: 8453,
  fromToken: '0x833589fCD6eDb6E08f4c7C32D4f71b54bdA02913',
  toToken: vaults[0].address, // vault address triggers Composer
  fromAmount: '100000000',
  fromAddress: '0xYourAddress', toAddress: '0xYourAddress',
});

Best Practices

  1. Always pass integrator — required for analytics, fee collection, and fee withdrawal.
  2. Re-fetch stale quotes — quotes expire in ~60s; always re-quote after an approval tx.
  3. Cache /chains, /tokens, /tools — they change infrequently and polling burns rate limit.
  4. Pass fromChain to /status — significantly faster responses.
  5. Use preset=stablecoin for stable-to-stable transfers instead of hand-tuning slippage/bridges.
  6. Request chainTypes explicitly on /chains if you need non-EVM (default is EVM-only).
  7. Amounts are strings in smallest units (human × 10^decimals); transactionRequest.value/gasLimit/gasPrice are hex strings.
  8. Handle PARTIAL completions — offer a same-chain recovery swap when worthwhile.
  9. For agents: consider the MCP server (https://mcp.li.quest/mcp) or @lifi/cli for token-efficient access; machine-readable docs at https://docs.li.fi/llms.txt and https://docs.li.fi/openapi.yaml (any docs page is fetchable as markdown by appending .md).

See references/REFERENCE.md for complete endpoint documentation, response schemas, the Earn and Intents API references, error codes, chain IDs, and token addresses.

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 Lifi AI skill do?

LI.FI REST API for cross-chain and same-chain token swaps, bridging, DeFi deposits (Composer), yield discovery (Earn), and intent-based execution (Intents). USE THIS SKILL WHEN USER WANTS TO: - Swap tokens between different blockchains (e.g., "swap USDC on Ethereum to ETH on Arbitrum") - Bridge tokens to another chain (e.g., "move my ETH from mainnet to Optimism") - Swap tokens on the same chain with best rates (e.g., "swap ETH to USDC on Polygon") - Find the best route or quote for a token swap across chains - Deposit into DeFi vaults/lending/staking in one click, including cross-chain (Co...

Why use Lifi on TypingMind?

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

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/internet-court/internet-court-skill/tree/main/vendored/lifi/lifi. 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 Lifi?

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 Lifi?

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

Is the Lifi AI skill free?

It is published on GitHub by internet-court. 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 👇