# LogicNodes (full reference)

> Gas prices for 14 EVM chains in the Blocknative format, plus read-only multi-chain and Aptos tools that never guess. MCP, A2A and REST. Free tier needs no key.

Machine-readable: https://logicnodes.io/openapi.json · https://logicnodes.io/llms.txt · https://logicnodes.io/.well-known/agent-card.json · https://logicnodes.io/docs
Identity: ERC-8004 agent #55092 on Base. Status: https://logicnodes.io/status · health: https://logicnodes.io/healthz

## Guarantees
- Existing endpoints keep the same responses (no auth, no payment) for the free tier.
- Fair use on the free tier: 120 requests/minute per IP; over the limit you get HTTP 429 with Retry-After.
- Stale data is an error (HTTP 503 JSON), never a silently old answer.

## Routes
- GET /gasprices/blockprices: Gas price estimates for 14 EVM chains, Blocknative-compatible (free)
- GET /v2/gas: Gas API v2: any EVM chain, multi-block horizon, confidence levels, USD cost (two-exchange cross-check), inputs echoed, receipt. Params: chain, confidenceLevels, blocks, usd (free)
- GET /v2/gas/history: Base fee and tip percentiles over the last 1-720 hours (one sample per minute, 30-day retention) (free)
- GET /v2/gas/accuracy: How the tip quoted at each confidence level compared with what blocks then contained (free)
- GET /docs/gas-method: Exactly how the estimates are computed, reproducible from the response (free)
- GET /a2a/health: Service health and per-chain data freshness (free)
- GET /healthz: Deep health for monitors: 200 only when every critical dependency is fine (free)
- GET /status: Human-readable status page (JSON at /status.json) (free)
- POST /mcp: MCP server (Streamable HTTP): get_gas_prices, list_supported_chains (free)
- POST /a2a: A2A JSON-RPC: message/send (free)
- GET /premium/snapshot: All-chains gas snapshot: Gas price estimates for every supported chain in a single call. (paid via x402)
- GET /premium/curve: Extended confidence curve: A dense gas-price curve (20 confidence points, 5-99%) for one chain, beyond the free tier's 12-level cap. (paid via x402)
- GET /premium/forecast: Next-blocks fee forecast: Deterministic worst/best-case base-fee bound for the next 1-10 blocks (EIP-1559's 12.5%-per-block cap), not a statistical prediction. (paid via x402)
- GET /v1/tools: List the 81 tools with their JSON Schemas (free)
- POST /paid/v1/tools/{name}: Pay-per-call version of the priced tools ($0.001-$0.005 USDC via x402). Errors are never charged (paid via x402)
- POST /v1/tools/{name}: Call a tool (JSON body = tool input). Same tools over MCP at /mcp/v2 (free)

## Gas endpoint
GET https://logicnodes.io/gasprices/blockprices?chainid=<id>&confidenceLevels=99,95,90,80,70
- chainid default 1. Supported: Ethereum (1), Base (8453), Arbitrum One (42161), OP Mainnet (10), Polygon PoS (137), BNB Chain (56), Avalanche C-Chain (43114), Linea (59144), Scroll (534352), zkSync Era (324), Gnosis (100), Blast (81457), Mantle (5000), Celo (42220).
- confidenceLevels: integers 1-99, up to 12, comma-separated or repeated.
- Response: { system, network, unit: "gwei", maxPrice, currentBlockNumber, msSinceLastBlock, blockPrices: [{ blockNumber, estimatedTransactionCount, baseFeePerGas, estimatedPrices: [{ confidence, price, maxPriorityFeePerGas, maxFeePerGas }] }] }
- Errors: 400 {error, message} for bad input; 503 when a chain's data is stale.

## Tools (REST https://logicnodes.io/v1/tools/<name> and MCP https://logicnodes.io/mcp/v2)
Every tool returns { tool, ok, value | error, provenance, receipt }. Chains are CAIP-2 ids only (eip155:8453, aptos:1, aptos:2). Ambiguous input returns AMBIGUOUS_INPUT with every option. Amounts are decimal strings.
### chain_info
Describes a supported chain: CAIP-2 id, display name, native currency, block explorers, and exactly which operations it supports. Omit `chain` to list every supported chain.
Input schema: {"type":"object","properties":{"chain":{"description":"CAIP-2 chain id such as eip155:8453 (Base) or aptos:1; a bare EVM chain id number also works. Chain names like \"base\" are refused.","type":["string","number"]}},"additionalProperties":false}

### gas_estimate
Current fee tiers for a chain as decimal strings with explicit units. EVM: confidence levels (default 99/95/90/80/70) in gwei with base fee, priority fee and max fee. Aptos: deprioritized/regular/prioritized gas unit prices in octas per gas unit. Data older than its freshness budget is an error, never a stale answer.
Input schema: {"type":"object","properties":{"chain":{"description":"CAIP-2 chain id such as eip155:8453 (Base) or aptos:1; a bare EVM chain id number also works. Chain names like \"base\" are refused.","type":["string","number"]},"confidenceLevels":{"description":"EVM only","minItems":1,"maxItems":12,"type":"array","items":{"type":"integer","minimum":1,"maximum":99}}},"required":["chain"],"additionalProperties":false}

### address_validate
Validates an address for a chain and returns its canonical form. EVM: EIP-55 (a wrong checksum is refused, never corrected). Aptos: AIP-40 strict form. Says only what can be known offline; it never guesses what kind of account an address is.
Input schema: {"type":"object","properties":{"chain":{"description":"CAIP-2 chain id such as eip155:8453 (Base) or aptos:1; a bare EVM chain id number also works. Chain names like \"base\" are refused.","type":["string","number"]},"address":{"type":"string"}},"required":["chain","address"],"additionalProperties":false}

### token_amount_convert
Converts between a human decimal amount and integer base units without ever rounding: too many decimals is an error. Decimals are never assumed: pass `decimals`, or an `asset` whose decimals are read from the chain (EVM: "native" or an ERC-20 token address, whose decimals() is read live from the chain; Aptos: any asset, with live decimals; an ambiguous symbol such as USDC returns AMBIGUOUS_INPUT with every option).
Input schema: {"type":"object","properties":{"chain":{"description":"CAIP-2 chain id such as eip155:8453 (Base) or aptos:1; a bare EVM chain id number also works. Chain names like \"base\" are refused.","type":["string","number"]},"amount":{"type":"string"},"direction":{"type":"string","enum":["toBase","toHuman"]},"decimals":{"type":"integer","minimum":0,"maximum":77},"asset":{"type":"string"}},"required":["chain","amount","direction"],"additionalProperties":false}

### tx_status
Whether a transaction is committed, failed, pending or not found, with a plain-English explanation. 'Not found' is reported as such and is not proof the transaction never existed.
Input schema: {"type":"object","properties":{"chain":{"description":"CAIP-2 chain id such as eip155:8453 (Base) or aptos:1; a bare EVM chain id number also works. Chain names like \"base\" are refused.","type":["string","number"]},"hash":{"type":"string"}},"required":["chain","hash"],"additionalProperties":false}

### aptos_address_validate
Validates an address against the Aptos address standard (AIP-40) and returns its canonical form: special addresses (0x0-0xf) as short form, all others as 0x + 64 digits. Strict by default; mode=relaxed also accepts a missing 0x or a few dropped leading zeros, and says which spelling you used (a truncated-looking address is always refused). With lookup=true it checks the chain to say whether the address is an account, an object, or neither, at one ledger version.
Input schema: {"type":"object","properties":{"address":{"type":"string"},"chain":{"description":"Aptos CAIP-2 id: aptos:1 (mainnet, default) or aptos:2 (testnet)","type":"string"},"mode":{"type":"string","enum":["strict","relaxed"]},"lookup":{"type":"boolean"}},"required":["address"],"additionalProperties":false}

### aptos_name_resolve
name -> address, or address -> primary name (give exactly one). A name resolves to its TARGET address, which can differ from its owner; a registered name with no target resolves to nothing (NOT_FOUND) and its owner is never substituted. An address's primary name is returned only if that name resolves forward to the same address, so a spoofed reverse record is never trusted.
Input schema: {"type":"object","properties":{"name":{"type":"string"},"address":{"type":"string"},"chain":{"description":"Aptos CAIP-2 id: aptos:1 (mainnet, default) or aptos:2 (testnet)","type":"string"}},"additionalProperties":false}

### aptos_asset_resolve
Turns a symbol, registry id, fungible-asset address or coin type into one asset, with its on-chain symbol/name/decimals read live. Several different assets share the symbols USDC and USDT (Circle/Tether native vs LayerZero-bridged); those return AMBIGUOUS_INPUT with every candidate, its issuer and how to select it. Assets outside the registry are returned with verified:false, never silently trusted.
Input schema: {"type":"object","properties":{"query":{"type":"string"},"chain":{"description":"Aptos CAIP-2 id: aptos:1 (mainnet, default) or aptos:2 (testnet)","type":"string"},"caseSensitive":{"type":"boolean"}},"required":["query"],"additionalProperties":false}

### aptos_amount_convert
Converts between a human decimal amount and integer base units, using the asset's decimals read LIVE from the chain (never assumed from a symbol). Refuses to round: more decimal places than the asset has is an error. An ambiguous symbol returns AMBIGUOUS_INPUT.
Input schema: {"type":"object","properties":{"amount":{"type":"string"},"direction":{"type":"string","enum":["toBase","toHuman"]},"asset":{"type":"string"},"chain":{"description":"Aptos CAIP-2 id: aptos:1 (mainnet, default) or aptos:2 (testnet)","type":"string"}},"required":["amount","direction","asset"],"additionalProperties":false}

### aptos_balance_get
Balances of an address for APT and fungible assets, all read at ONE ledger version (stated, or the current one pinned). Amounts are exact integer strings plus decimal strings. Default assets: APT and every verified asset; pass `assets` (symbols, ids, addresses, coin types) to choose. An ambiguous symbol fails the whole call. An address with no account and all-zero balances is reported honestly, not as an empty account.
Input schema: {"type":"object","properties":{"address":{"type":"string"},"assets":{"minItems":1,"maxItems":20,"type":"array","items":{"type":"string"}},"ledgerVersion":{"description":"read the chain exactly as of this ledger version (default: the current one, pinned for consistency)","anyOf":[{"type":"string"},{"type":"integer","minimum":0,"maximum":9007199254740991}]},"chain":{"description":"Aptos CAIP-2 id: aptos:1 (mainnet, default) or aptos:2 (testnet)","type":"string"}},"required":["address"],"additionalProperties":false}

### aptos_fee_estimate
Gas unit price tiers (deprioritized/regular/prioritized) plus the cost of common transactions in octas, APT and USD. Gas units are MEASURED by simulating real transfers now, not taken from a table. USD comes only from two independent exchanges that agree; otherwise USD is explicitly unavailable. Transaction types: apt_transfer, apt_transfer_new_account; for anything else use aptos_tx_simulate.
Input schema: {"type":"object","properties":{"txTypes":{"type":"array","items":{"type":"string"}},"chain":{"description":"Aptos CAIP-2 id: aptos:1 (mainnet, default) or aptos:2 (testnet)","type":"string"}},"additionalProperties":false}

### aptos_tx_simulate
Runs an UNSIGNED entry-function transaction through the node's own simulator (no key needed, nothing is submitted) and returns success or the abort decoded in plain English, exact gas and fee, balance changes, and events. Arguments are type-checked against the module ABI first. Balance changes count only the framework's Withdraw/Deposit events, so assets that re-emit them (like USDC) are never double-counted. A simulation runs against the node's current state and cannot be pinned to an older ledger version.
Input schema: {"type":"object","properties":{"sender":{"type":"string"},"function":{"type":"string"},"typeArguments":{"type":"array","items":{"type":"string"}},"arguments":{"type":"array","items":{}},"maxGasAmount":{"anyOf":[{"type":"string"},{"type":"integer","minimum":0,"maximum":9007199254740991}]},"gasUnitPrice":{"anyOf":[{"type":"string"},{"type":"integer","minimum":0,"maximum":9007199254740991}]},"sequenceNumber":{"anyOf":[{"type":"string"},{"type":"integer","minimum":0,"maximum":9007199254740991}]},"chain":{"description":"Aptos CAIP-2 id: aptos:1 (mainnet, default) or aptos:2 (testnet)","type":"string"}},"required":["sender","function"],"additionalProperties":false}

### aptos_tx_decode
Explains what an entry-function payload would do, in plain English, with every argument type-checked against the module ABI and amounts shown using the asset's live decimals. Known patterns (transfers, batch transfers, key rotation, code publishing) carry a risk flag from a fixed table; no flag means 'not recognised', never 'safe'.
Input schema: {"type":"object","properties":{"function":{"type":"string"},"typeArguments":{"type":"array","items":{"type":"string"}},"arguments":{"type":"array","items":{}},"chain":{"description":"Aptos CAIP-2 id: aptos:1 (mainnet, default) or aptos:2 (testnet)","type":"string"}},"required":["function"],"additionalProperties":false}

### aptos_tx_status
Committed, failed, pending or not found for a transaction hash, with the VM status explained in plain English (Move aborts decoded to module, error name, code and category; the raw status is always included). Fee is computed from the VM's own fee statement.
Input schema: {"type":"object","properties":{"hash":{"type":"string"},"chain":{"description":"Aptos CAIP-2 id: aptos:1 (mainnet, default) or aptos:2 (testnet)","type":"string"}},"required":["hash"],"additionalProperties":false}

### aptos_view
Calls a Move #[view] function read-only, after checking every type argument and argument against the module's ABI: u64/u128/u256 must be decimal strings, vector<u8> a 0x hex string, Option<T> written {"vec":[]} or {"vec":[value]}, addresses in AIP-40 form. A wrong argument is a specific error from us, never an opaque node error. Optionally reads at a stated ledger version.
Input schema: {"type":"object","properties":{"function":{"type":"string"},"typeArguments":{"type":"array","items":{"type":"string"}},"arguments":{"type":"array","items":{}},"ledgerVersion":{"description":"read the chain exactly as of this ledger version (default: the current one, pinned for consistency)","anyOf":[{"type":"string"},{"type":"integer","minimum":0,"maximum":9007199254740991}]},"chain":{"description":"Aptos CAIP-2 id: aptos:1 (mainnet, default) or aptos:2 (testnet)","type":"string"}},"required":["function"],"additionalProperties":false}

### decimal_calc
Add, subtract, multiply, divide, raise to an integer power, compare, or take a percentage, on decimal STRINGS with exact arithmetic (no floating point). The result is rounded only to the scale and rule you choose, and the response says whether rounding changed the value. JSON numbers are refused because they lose precision. Division by zero is an error.
Input schema: {"type":"object","properties":{"op":{"type":"string","enum":["add","subtract","multiply","divide","power","compare","percent_of"]},"a":{"type":"string"},"b":{"type":"string"},"scale":{"type":"integer","minimum":0,"maximum":36},"rounding":{"description":"rounding rule; default half_up (ties away from zero). half_even is banker's rounding","type":"string","enum":["down","up","floor","ceil","half_up","half_even","half_down"]}},"required":["op","a","b"],"additionalProperties":false}

### table_calc
Give it your rows (JSON or CSV) and the figures you need; it returns them computed exactly (decimal arithmetic, no floating point), so your agent never has to add up a column in its head. Metrics are expressions: sum([Amount]), avg(price), count(), count_distinct([Customer]), median(days), min/max, sum(qty * price), sum(amount) / total(amount) * 100 (share of the overall total), round(x, 2), avg(days_between([Invoiced], [Paid])), sum(age_days([Due])) with asOf. Group by columns or periods: month([Date]), week, quarter, year, weekday, or ranges such as bucket(age_days([Due]), 30, 60, 90) for an aging report. Optional: where filters (numbers or dates), sort, limit. Cells that are not numbers are never counted as 0: they are skipped and reported with row numbers (or refused with strictNumbers). Up to 10000 rows. The same input always gives the same output, with a receipt.
Input schema: {"type":"object","properties":{"rows":{"description":"the data as JSON objects, one per row (up to 10000). Numbers as decimal strings are safest; JSON numbers are accepted","maxItems":10000,"type":"array","items":{"type":"object","propertyNames":{"type":"string","maxLength":200},"additionalProperties":{"anyOf":[{"type":"string","maxLength":10000},{"type":"number"},{"type":"boolean"},{"type":"null"}]}}},"csv":{"description":"the data as CSV text with a header row (instead of rows)","type":"string","maxLength":900000},"delimiter":{"description":"CSV delimiter, default comma","type":"string","enum":[",",";","\t","|"]},"where":{"description":"keep only rows matching ALL of these: eq, ne, gt, gte, lt, lte (numeric when both sides are numbers), contains (case-insensitive), in (list), empty, not_empty","maxItems":20,"type":"array","items":{"type":"object","properties":{"column":{"type":"string","minLength":1,"maxLength":200},"op":{"type":"string","enum":["eq","ne","gt","gte","lt","lte","contains","in","empty","not_empty"]},"value":{"description":"text; a list for \"in\"; numbers as decimal strings","anyOf":[{"type":"string","maxLength":500},{"maxItems":200,"type":"array","items":{"type":"string","maxLength":500}}]}},"required":["column","op"],"additionalProperties":false}},"strictNumbers":{"description":"true: refuse any cell that is not a plain decimal (no $, commas, %, or text in number columns). Default false: those are read and every change is reported","type":"boolean"},"dateFormat":{"description":"how to read slash dates like 03/04/2026: mdy (US) or dmy; ISO dates (2026-03-04, with or without a time) are always read. Default iso","type":"string","enum":["iso","mdy","dmy"]},"asOf":{"description":"the date age_days() counts up to, YYYY-MM-DD (required for age_days: answers never silently depend on today's date)","type":"string","pattern":"^\\d{4}-\\d{2}-\\d{2}$"},"metrics":{"minItems":1,"maxItems":20,"type":"array","items":{"type":"object","properties":{"name":{"type":"string","minLength":1,"maxLength":80,"description":"label for this figure"},"expr":{"type":"string","minLength":1,"maxLength":500,"description":"e.g. sum([Amount]) or sum(qty * price) / count()"}},"required":["name","expr"],"additionalProperties":false}},"groupBy":{"description":"one figure set per distinct value: a column name, or a key such as month([Date]), week([Date]) (ISO), quarter, year, day, weekday, or bucket(age_days([Due]), 30, 60, 90) for ranges","maxItems":3,"type":"array","items":{"type":"string","minLength":1,"maxLength":200}},"sort":{"type":"object","properties":{"by":{"type":"string","description":"a metric name or a groupBy column"},"dir":{"type":"string","enum":["asc","desc"]}},"required":["by"],"additionalProperties":false},"limit":{"description":"return at most this many groups","type":"integer","minimum":1,"maximum":1000},"scale":{"description":"decimal places for every figure; default: exact, or 18 places when a value has no finite decimal form (like 1/3)","type":"integer","minimum":0,"maximum":18},"rounding":{"description":"rounding rule when scale applies; default half_up","type":"string","enum":["down","up","floor","ceil","half_up","half_even","half_down"]}},"required":["metrics"],"additionalProperties":false}

### verify_answer
Before your agent sends a figure to a customer, a ledger or another agent, check it. Each claim is an expression and the value you believe it equals: plain arithmetic ("1299.99 * 3 * 1.0825" expected "4221.87") or over your rows ("sum([Amount])" expected "4312.50", with optional per-claim where filters). Answers per claim: "correct" (exactly equal, or within your tolerance), "correct_when_rounded" (equal once the true value is rounded half-up to the decimals you wrote), or "wrong", always with the exact value and the difference. Same exact engine as table_calc and decimal_calc; up to 50 claims per call, with a receipt you can attach as proof.
Input schema: {"type":"object","properties":{"rows":{"description":"the data as JSON objects, one per row (up to 10000). Numbers as decimal strings are safest; JSON numbers are accepted","maxItems":10000,"type":"array","items":{"type":"object","propertyNames":{"type":"string","maxLength":200},"additionalProperties":{"anyOf":[{"type":"string","maxLength":10000},{"type":"number"},{"type":"boolean"},{"type":"null"}]}}},"csv":{"description":"the data as CSV text with a header row (instead of rows)","type":"string","maxLength":900000},"delimiter":{"description":"CSV delimiter, default comma","type":"string","enum":[",",";","\t","|"]},"where":{"description":"keep only rows matching ALL of these: eq, ne, gt, gte, lt, lte (numeric when both sides are numbers), contains (case-insensitive), in (list), empty, not_empty","maxItems":20,"type":"array","items":{"type":"object","properties":{"column":{"type":"string","minLength":1,"maxLength":200},"op":{"type":"string","enum":["eq","ne","gt","gte","lt","lte","contains","in","empty","not_empty"]},"value":{"description":"text; a list for \"in\"; numbers as decimal strings","anyOf":[{"type":"string","maxLength":500},{"maxItems":200,"type":"array","items":{"type":"string","maxLength":500}}]}},"required":["column","op"],"additionalProperties":false}},"strictNumbers":{"description":"true: refuse any cell that is not a plain decimal (no $, commas, %, or text in number columns). Default false: those are read and every change is reported","type":"boolean"},"dateFormat":{"description":"how to read slash dates like 03/04/2026: mdy (US) or dmy; ISO dates (2026-03-04, with or without a time) are always read. Default iso","type":"string","enum":["iso","mdy","dmy"]},"asOf":{"description":"the date age_days() counts up to, YYYY-MM-DD (required for age_days: answers never silently depend on today's date)","type":"string","pattern":"^\\d{4}-\\d{2}-\\d{2}$"},"claims":{"minItems":1,"maxItems":50,"type":"array","items":{"type":"object","properties":{"label":{"description":"what the figure is, e.g. \"invoice total\"","type":"string","maxLength":120},"expr":{"type":"string","minLength":1,"maxLength":500,"description":"arithmetic, or an aggregate over the rows: sum([Amount]), count(), avg(price), sum(qty * price) * 1.08"},"expected":{"type":"string","maxLength":100,"description":"the value to check, as a decimal string; $, commas and % are read"},"tolerance":{"description":"accept a difference up to this (decimal string, e.g. \"0.01\"); default 0","type":"string","maxLength":40},"where":{"description":"filters for this claim only, on top of the top-level where","maxItems":20,"type":"array","items":{"type":"object","properties":{"column":{"type":"string","minLength":1,"maxLength":200},"op":{"type":"string","enum":["eq","ne","gt","gte","lt","lte","contains","in","empty","not_empty"]},"value":{"description":"text; a list for \"in\"; numbers as decimal strings","anyOf":[{"type":"string","maxLength":500},{"maxItems":200,"type":"array","items":{"type":"string","maxLength":500}}]}},"required":["column","op"],"additionalProperties":false}}},"required":["expr","expected"],"additionalProperties":false}}},"required":["claims"],"additionalProperties":false}

### quote_check
Before your agent quotes, cites or summarises a document, check each quote against the source text. Per quote: "exact" (verbatim, with location), "same_words" (only capitalisation, spacing or punctuation differ), "numbers_differ" (the passage exists but a number was changed, e.g. 12% vs 21%), "close_but_different" (a similar passage exists, with the words that differ), or "not_found" (the source does not say it). Also flags a negation that was added or dropped ("not"). Returns the line, character offsets and the source passage. Up to 50 quotes per call against up to 600000 characters of source. Deterministic string matching, no model.
Input schema: {"type":"object","properties":{"source":{"type":"string","minLength":1,"maxLength":600000,"description":"the full source text (document, page, transcript)"},"quotes":{"minItems":1,"maxItems":50,"type":"array","items":{"type":"object","properties":{"text":{"type":"string","minLength":1,"maxLength":5000,"description":"the quote or claimed wording"},"label":{"type":"string","maxLength":120}},"required":["text"],"additionalProperties":false}}},"required":["source","quotes"],"additionalProperties":false}

### text_facts
Facts about a text that language models routinely get wrong: characters (as people see them, emoji included), words, lines, paragraphs, UTF-8 bytes, SMS segments (GSM-7 or UCS-2) and how often a word or letter appears (with positions; e.g. how many r in strawberry). Checks limits you give (max characters, words or SMS segments) before you post or send. With compareTo it returns exactly what changed between two versions (word or line diff). Deterministic, no model.
Input schema: {"type":"object","properties":{"text":{"type":"string","maxLength":600000},"count":{"description":"words, phrases or letters to count","maxItems":50,"type":"array","items":{"type":"object","properties":{"term":{"type":"string","minLength":1,"maxLength":200},"caseSensitive":{"description":"default false","type":"boolean"},"wholeWord":{"description":"default false: also counts inside longer words","type":"boolean"}},"required":["term"],"additionalProperties":false}},"limits":{"type":"object","properties":{"maxCharacters":{"type":"integer","minimum":0,"maximum":9007199254740991},"maxWords":{"type":"integer","minimum":0,"maximum":9007199254740991},"maxSmsSegments":{"type":"integer","minimum":0,"maximum":9007199254740991}},"additionalProperties":false},"compareTo":{"description":"another version of the text: returns what changed from compareTo to text","type":"string","maxLength":600000},"diffBy":{"description":"diff granularity, default word","type":"string","enum":["word","line"]}},"required":["text"],"additionalProperties":false}

### reconcile
Match two lists that should agree, line by line, and list everything that does not: bank statement vs ledger, invoices vs payments received, payouts vs orders. Rules, in order: same reference and amount; same reference but a different amount (reported as a mismatch); then same amount within amountTolerance and dates within dateWindowDays. Each line matches at most once. Set sign "opposite" when one side records money out as negative, or "ignore" to compare sizes only. Returns matched pairs (row numbers on both sides), amount mismatches, unmatched lines on each side, and exact totals: left, right, matched, unmatched and the difference. Up to 5000 lines per side. Deterministic, exact decimals.
Input schema: {"type":"object","properties":{"left":{"type":"object","properties":{"rows":{"description":"the lines as JSON objects","maxItems":5000,"type":"array","items":{"type":"object","propertyNames":{"type":"string","maxLength":200},"additionalProperties":{"anyOf":[{"type":"string","maxLength":2000},{"type":"number"},{"type":"boolean"},{"type":"null"}]}}},"csv":{"description":"the lines as CSV with a header row (instead of rows)","type":"string","maxLength":450000},"delimiter":{"type":"string","enum":[",",";","\t","|"]},"amount":{"type":"string","minLength":1,"maxLength":200,"description":"the amount column"},"date":{"description":"the date column (optional; enables the date window)","type":"string","maxLength":200},"ref":{"description":"a reference column to match on first: invoice number, order id, transaction id","type":"string","maxLength":200},"name":{"description":"what this list is, e.g. \"bank\" or \"ledger\"","type":"string","maxLength":60}},"required":["amount"],"additionalProperties":false,"description":"the first list, e.g. the bank statement"},"right":{"type":"object","properties":{"rows":{"description":"the lines as JSON objects","maxItems":5000,"type":"array","items":{"type":"object","propertyNames":{"type":"string","maxLength":200},"additionalProperties":{"anyOf":[{"type":"string","maxLength":2000},{"type":"number"},{"type":"boolean"},{"type":"null"}]}}},"csv":{"description":"the lines as CSV with a header row (instead of rows)","type":"string","maxLength":450000},"delimiter":{"type":"string","enum":[",",";","\t","|"]},"amount":{"type":"string","minLength":1,"maxLength":200,"description":"the amount column"},"date":{"description":"the date column (optional; enables the date window)","type":"string","maxLength":200},"ref":{"description":"a reference column to match on first: invoice number, order id, transaction id","type":"string","maxLength":200},"name":{"description":"what this list is, e.g. \"bank\" or \"ledger\"","type":"string","maxLength":60}},"required":["amount"],"additionalProperties":false,"description":"the second list, e.g. the ledger"},"amountTolerance":{"description":"largest amount difference that still counts as equal, e.g. \"0.01\"; default 0 (exact)","type":"string","maxLength":40},"dateWindowDays":{"description":"largest date gap for a match when both lists have dates; default 3 (bank posting delays)","type":"integer","minimum":0,"maximum":90},"sign":{"description":"same (default): amounts compared as written; opposite: right-side amounts are negated first; ignore: compare absolute values","type":"string","enum":["same","opposite","ignore"]},"dateFormat":{"description":"how to read slash dates; ISO dates always work","type":"string","enum":["iso","mdy","dmy"]}},"required":["left","right"],"additionalProperties":false}

### json_check
Check structured output before you use or send it. Give the raw text (a model reply works: set extract to pull the JSON out of ```json fences or surrounding prose) or the parsed data, and optionally a JSON Schema. Returns valid / invalid with every problem and its JSON Pointer path (e.g. /items/2/price: must be >= 0), or the line and column of a syntax error. Supports type, enum, const, required, properties, additionalProperties, patternProperties, items/prefixItems, min/max (length, items, properties), uniqueItems, contains, minimum/maximum/exclusive, multipleOf (exact decimals), pattern, format (email, date, date-time, time, uri, uuid, ipv4, ipv6, hostname) asserted, allOf/anyOf/oneOf/not, if/then/else and local $ref. Up to 100 errors listed.
Input schema: {"type":"object","properties":{"text":{"description":"the JSON as text (or a model reply containing it, with extract: true)","type":"string","maxLength":500000},"data":{"description":"the value itself, already parsed (instead of text)"},"schema":{"description":"a JSON Schema; omit to only check that the text is valid JSON","anyOf":[{"type":"object","propertyNames":{"type":"string"},"additionalProperties":{}},{"type":"boolean"}]},"extract":{"description":"pull the JSON out of surrounding text: a ```json block, else the first {...} or [...]","type":"boolean"}},"additionalProperties":false}

### invoice_check
Line totals (qty x unit price, minus line discounts), subtotal, invoice discount (spread over the lines by value), tax per invoice or per line (the usual source of 1-cent differences: both are shown), shipping (taxable or not) and the total, in exact decimals rounded to cents only where an invoice rounds. Pass the figures written on an invoice in claimed (and claimedTotal per line) and each comes back correct or wrong with the right value. You choose the rates; not tax advice.
Input schema: {"type":"object","properties":{"lines":{"minItems":1,"maxItems":500,"type":"array","items":{"type":"object","properties":{"description":{"type":"string","maxLength":300},"qty":{"type":"string","maxLength":40,"description":"quantity as a decimal string, e.g. \"3\" or \"1.5\""},"unitPrice":{"type":"string","maxLength":40,"description":"price per unit, e.g. \"19.99\""},"discountPercent":{"description":"line discount in percent, e.g. \"10\"","type":"string","maxLength":20},"discountAmount":{"description":"line discount as an amount","type":"string","maxLength":40},"taxable":{"description":"default true","type":"boolean"},"taxRatePercent":{"description":"this line's tax rate if it differs from the invoice rate","type":"string","maxLength":20},"claimedTotal":{"description":"the line total written on the invoice, to check","type":"string","maxLength":40}},"required":["qty","unitPrice"],"additionalProperties":false}},"taxRatePercent":{"description":"e.g. \"8.25\"; default 0","type":"string","maxLength":20},"taxRounding":{"description":"round tax once on the invoice (default) or on each line and add up","type":"string","enum":["per_invoice","per_line"]},"discountPercent":{"description":"invoice-level discount in percent, before tax","type":"string","maxLength":20},"discountAmount":{"description":"invoice-level discount amount, before tax","type":"string","maxLength":40},"shipping":{"type":"string","maxLength":40},"shippingTaxable":{"description":"default false","type":"boolean"},"scale":{"description":"decimal places of the currency, default 2","type":"integer","minimum":0,"maximum":6},"rounding":{"description":"default half_up","type":"string","enum":["down","up","floor","ceil","half_up","half_even","half_down"]},"claimed":{"description":"figures written on the invoice, to check","type":"object","properties":{"subtotal":{"type":"string","maxLength":40},"discount":{"type":"string","maxLength":40},"tax":{"type":"string","maxLength":40},"shipping":{"type":"string","maxLength":40},"total":{"type":"string","maxLength":40}},"additionalProperties":false}},"required":["lines"],"additionalProperties":false}

### timesheet_calc
From clock-in/clock-out entries per worker: hours worked (breaks removed, overnight shifts handled), optional punch rounding (e.g. to the nearest 15 minutes), regular / overtime / double-time hours under the rules you set (daily after N hours, double after M, weekly after 40, applied without counting an hour twice), and pay at each entry's rate. Per worker and per entry, with warnings for overlapping or very long shifts. Minutes are counted exactly; pay is rounded to cents once per worker. You choose the rules; not legal advice.
Input schema: {"type":"object","properties":{"entries":{"minItems":1,"maxItems":5000,"type":"array","items":{"type":"object","properties":{"worker":{"type":"string","minLength":1,"maxLength":120},"date":{"description":"YYYY-MM-DD (needed for weekly overtime)","type":"string","maxLength":40},"start":{"type":"string","maxLength":20,"description":"clock-in, e.g. \"08:00\" or \"8:00 AM\""},"end":{"type":"string","maxLength":20,"description":"clock-out; earlier than start means the next day (overnight)"},"breakMinutes":{"description":"unpaid break minutes","type":"integer","minimum":0,"maximum":720},"rate":{"description":"hourly rate for this entry, e.g. \"22.50\"","type":"string","maxLength":40}},"required":["worker","start","end"],"additionalProperties":false}},"defaultRate":{"description":"hourly rate when an entry has none","type":"string","maxLength":40},"roundPunchesTo":{"description":"round each clock-in and clock-out to the nearest N minutes (ties round up)","type":"integer","minimum":1,"maximum":60},"overtime":{"type":"object","properties":{"dailyAfterHours":{"description":"hours per day before overtime, e.g. 8 (California)","type":"number","minimum":0,"maximum":24},"doubleAfterDailyHours":{"description":"hours per day before double time, e.g. 12","type":"number","minimum":0,"maximum":24},"weeklyAfterHours":{"description":"hours per week before overtime, e.g. 40 (US federal)","type":"number","minimum":0,"maximum":168},"multiplier":{"description":"overtime pay multiplier, default \"1.5\"","type":"string","maxLength":10},"doubleMultiplier":{"description":"double-time multiplier, default \"2\"","type":"string","maxLength":10}},"additionalProperties":false},"weekStarts":{"description":"first day of the pay week, default monday","type":"string","enum":["monday","sunday"]}},"required":["entries"],"additionalProperties":false}

### list_ops
Set and order operations on lists of up to 50000 items, done exactly (language models drop or duplicate items in long lists). ops: dedupe (with the duplicates and their counts), sort (natural order: item2 before item10), union, intersection, difference (in a, not in b), symmetric_difference, frequency (each value and how often), compare (only in a, only in b, in both). Options: ignoreCase, trim (default true), sort the result. Order is first appearance unless you sort. Deterministic.
Input schema: {"type":"object","properties":{"op":{"type":"string","enum":["dedupe","sort","union","intersection","difference","symmetric_difference","frequency","compare"]},"a":{"maxItems":50000,"type":"array","items":{"type":"string","maxLength":1000},"description":"the first list"},"b":{"description":"the second list (union, intersection, difference, symmetric_difference, compare)","maxItems":50000,"type":"array","items":{"type":"string","maxLength":1000}},"ignoreCase":{"description":"treat \"Acme\" and \"ACME\" as the same (the first spelling seen is kept); default false","type":"boolean"},"trim":{"description":"remove leading/trailing spaces and empty items first; default true","type":"boolean"},"sort":{"description":"order of the result; default none (first appearance), asc for op sort","type":"string","enum":["none","asc","desc"]},"natural":{"description":"natural order when sorting (item2 < item10); default true","type":"boolean"}},"required":["op","a"],"additionalProperties":false}

### contact_normalize
Normalise contact lists before they go into a CRM or a mail/SMS tool: emails (trimmed, display names and mailto: removed, domain lower-cased, syntax checked, typo domains like gmial.com flagged with a suggestion, role inboxes and disposable domains flagged) and phone numbers to E.164 (+14805550123; North American numbers checked against the numbering plan, others by shape, extensions kept). Duplicates that differ only in formatting are grouped. Up to 10000 of each. Deterministic, no lookups.
Input schema: {"type":"object","properties":{"emails":{"maxItems":10000,"type":"array","items":{"type":"string","maxLength":320}},"phones":{"maxItems":10000,"type":"array","items":{"type":"string","maxLength":64}},"defaultCallingCode":{"description":"country calling code for numbers written without one, e.g. \"1\" (US/Canada, default) or \"44\" (UK)","type":"string","pattern":"^[1-9]\\d{0,2}$"}},"additionalProperties":false}

### find_free_slots
Finds meeting slots inside every participant's working hours (each in their own IANA time zone, daylight-saving changes included) that avoid all their busy times (optionally with a buffer). Returns each slot in your display zone and in every participant's local time, so nobody has to convert. Up to 31 days, 20 participants. Deterministic: zone rules come from the IANA database, not a guessed offset.
Input schema: {"type":"object","properties":{"participants":{"minItems":1,"maxItems":20,"type":"array","items":{"type":"object","properties":{"name":{"type":"string","minLength":1,"maxLength":120},"timeZone":{"type":"string","maxLength":64,"description":"IANA zone, e.g. America/Phoenix, Europe/London"},"workingHours":{"description":"default 09:00-17:00, Monday to Friday, in their own zone","type":"object","properties":{"start":{"type":"string","maxLength":10,"description":"e.g. 09:00"},"end":{"type":"string","maxLength":10,"description":"e.g. 17:00"},"days":{"description":"default mon-fri","minItems":1,"maxItems":7,"type":"array","items":{"type":"string","enum":["mon","tue","wed","thu","fri","sat","sun"]}}},"required":["start","end"],"additionalProperties":false},"busy":{"description":"busy times as instants with an offset, e.g. 2026-11-02T15:00:00Z or 2026-11-02T08:00:00-07:00","maxItems":500,"type":"array","items":{"type":"object","properties":{"start":{"type":"string","maxLength":40},"end":{"type":"string","maxLength":40}},"required":["start","end"],"additionalProperties":false}}},"required":["name","timeZone"],"additionalProperties":false}},"from":{"type":"string","pattern":"^\\d{4}-\\d{2}-\\d{2}$","description":"first date to search, YYYY-MM-DD, in the display zone"},"to":{"type":"string","pattern":"^\\d{4}-\\d{2}-\\d{2}$","description":"last date to search (inclusive)"},"durationMinutes":{"type":"integer","minimum":5,"maximum":720},"stepMinutes":{"description":"start times on this grid of the display zone's clock, default 30","type":"integer","minimum":5,"maximum":120},"bufferMinutes":{"description":"keep this much free before and after busy times","type":"integer","minimum":0,"maximum":120},"displayTimeZone":{"description":"zone for from/to and for listing slots; default the first participant's","type":"string","maxLength":64},"maxResults":{"description":"default 20","type":"integer","minimum":1,"maximum":200}},"required":["participants","from","to","durationMinutes"],"additionalProperties":false}

### date_series
Lists the dates of a recurrence exactly: daily / weekly (on chosen weekdays) / monthly (on day numbers like 1, 15 or -1 for the last day, or the nth weekday like 2nd Tuesday or -1 for the last Friday) / yearly, every N periods, from a start date up to a count or an until date. Days that do not exist (31 February) are skipped or clamped as you choose. Optional business-day adjustment (following, preceding, modified following) with your weekend rule and holidays, each moved date marked. Up to 500 dates.
Input schema: {"type":"object","properties":{"start":{"type":"string","pattern":"^\\d{4}-\\d{2}-\\d{2}$","description":"first possible date, YYYY-MM-DD"},"freq":{"type":"string","enum":["daily","weekly","monthly","yearly"]},"interval":{"description":"every N periods, default 1","type":"integer","minimum":1,"maximum":366},"count":{"description":"how many dates (default 12 when no until)","type":"integer","minimum":1,"maximum":500},"until":{"description":"last possible date (inclusive)","type":"string","pattern":"^\\d{4}-\\d{2}-\\d{2}$"},"weekdays":{"description":"weekly: which days; monthly with nth: the weekday","minItems":1,"maxItems":7,"type":"array","items":{"type":"string","enum":["mon","tue","wed","thu","fri","sat","sun"]}},"nth":{"description":"monthly: 1 = first, 2 = second, -1 = last of the weekday in weekdays","type":"integer","minimum":-5,"maximum":5},"monthDays":{"description":"monthly: day numbers, -1 = last day of the month","minItems":1,"maxItems":31,"type":"array","items":{"type":"integer","minimum":-31,"maximum":31}},"missingDay":{"description":"when a month lacks the day (31st, Feb 29): skip (default) or use the last day","type":"string","enum":["skip","clamp"]},"adjust":{"description":"move dates off weekends/holidays; modified_following stays in the same month","type":"string","enum":["none","following","preceding","modified_following"]},"weekend":{"type":"string","enum":["sat_sun","fri_sat","sun_only","none"]},"holidays":{"maxItems":400,"type":"array","items":{"type":"string","pattern":"^\\d{4}-\\d{2}-\\d{2}$"}}},"required":["start","freq"],"additionalProperties":false}

### table_join
Bring columns from one table into another by a shared key (customer id, email, SKU), exactly: how "left" keeps every left row (blank when no match), "inner" keeps matches only, "anti" lists left rows with no match. Keys (up to 3 columns) match exactly, ignoring case, or as numbers ("00123" equals "123"). When the right table has the key more than once, take the first (like VLOOKUP) or all matches; duplicates are reported either way. Returns the joined rows plus counts of matched and unmatched rows on both sides. Up to 10000 rows per table.
Input schema: {"type":"object","properties":{"left":{"type":"object","properties":{"rows":{"maxItems":10000,"type":"array","items":{"type":"object","propertyNames":{"type":"string","maxLength":200},"additionalProperties":{"anyOf":[{"type":"string","maxLength":5000},{"type":"number"},{"type":"boolean"},{"type":"null"}]}}},"csv":{"type":"string","maxLength":450000},"delimiter":{"type":"string","enum":[",",";","\t","|"]}},"additionalProperties":false,"description":"the table to enrich"},"right":{"type":"object","properties":{"rows":{"maxItems":10000,"type":"array","items":{"type":"object","propertyNames":{"type":"string","maxLength":200},"additionalProperties":{"anyOf":[{"type":"string","maxLength":5000},{"type":"number"},{"type":"boolean"},{"type":"null"}]}}},"csv":{"type":"string","maxLength":450000},"delimiter":{"type":"string","enum":[",",";","\t","|"]}},"additionalProperties":false,"description":"the lookup table"},"on":{"minItems":1,"maxItems":3,"type":"array","items":{"type":"object","properties":{"left":{"type":"string","minLength":1,"maxLength":200},"right":{"type":"string","minLength":1,"maxLength":200}},"required":["left","right"],"additionalProperties":false},"description":"key columns, e.g. [{left: \"Email\", right: \"email\"}]"},"how":{"description":"default left","type":"string","enum":["left","inner","anti"]},"match":{"description":"how keys compare; default exact (after trimming spaces)","type":"string","enum":["exact","ignore_case","numeric"]},"duplicates":{"description":"right-side keys that repeat: first match (default, like VLOOKUP) or every match","type":"string","enum":["first","all"]},"columns":{"description":"right-side columns to bring in (default all except the keys)","maxItems":100,"type":"array","items":{"type":"string","maxLength":200}}},"required":["left","right","on"],"additionalProperties":false}

### pii_redact
Scans text for emails, phone numbers, US SSNs, payment card numbers (Luhn-checked), IBANs (checksum-checked), IP addresses and secrets (API keys such as sk-..., AKIA..., ghp_..., xoxb-..., private keys and 0x-prefixed 64-hex keys), and returns the text redacted: label mode gives the same value the same placeholder ([EMAIL_1]) so the text still reads, mask keeps the last 4 characters, remove deletes. Findings list type, position and a masked preview, never the full value. Pattern and checksum based: it will not catch names or street addresses.
Input schema: {"type":"object","properties":{"text":{"type":"string","maxLength":600000},"types":{"description":"which kinds to find, default all","minItems":1,"maxItems":7,"type":"array","items":{"type":"string","enum":["email","phone","ssn","card","iban","ip","secret"]}},"mode":{"description":"default label","type":"string","enum":["label","mask","remove"]}},"required":["text"],"additionalProperties":false}

### id_check
Catches typos in identifiers before they go on an invoice, a payment or an order: IBAN (country length + mod-97), US ABA routing number (3-7-1 checksum), GTIN/EAN/UPC barcodes, ISBN-10/13, VIN (check digit), US NPI, EIN (format and IRS prefix), or any Luhn number. type auto guesses from the shape. Up to 500 per call. A valid checksum means well-formed, not that the account or product exists.
Input schema: {"type":"object","properties":{"ids":{"minItems":1,"maxItems":500,"type":"array","items":{"type":"object","properties":{"value":{"type":"string","minLength":1,"maxLength":64},"type":{"type":"string","enum":["auto","iban","gtin","isbn","aba_routing","vin","npi","luhn","ein"]}},"required":["value"],"additionalProperties":false}}},"required":["ids"],"additionalProperties":false}

### table_profile
Reads every cell of a table (JSON rows or CSV, up to 20000 rows) and reports per column: type (integer, decimal, date, boolean, text, mixed), filled and blank counts, distinct values, most common values, number sum/min/max (exact), date range, and whether it is unique (a possible id). Lists the problems that break later steps: numbers written with $ or commas, mixed date formats, ambiguous slash dates, stray spaces, the same value in different capitalisation, and duplicate rows (with row numbers). Use it before table_calc or table_join on data you have not seen.
Input schema: {"type":"object","properties":{"rows":{"maxItems":20000,"type":"array","items":{"type":"object","propertyNames":{"type":"string","maxLength":200},"additionalProperties":{"anyOf":[{"type":"string","maxLength":5000},{"type":"number"},{"type":"boolean"},{"type":"null"}]}}},"csv":{"type":"string","maxLength":900000},"delimiter":{"type":"string","enum":[",",";","\t","|"]}},"additionalProperties":false}

### holiday_calendar
Computes public holidays from the rules in law instead of memory: US federal holidays (fixed dates moved to Friday/Monday when they fall on a weekend, nth-weekday holidays such as Thanksgiving, Juneteenth from 2021) or bank holidays in England and Wales (Easter-based, substitute days for weekends, Christmas/Boxing Day combinations, and the one-off royal holidays proclaimed so far). For each: name, the calendar date, the observed day off and weekday. Also returns the observed dates as a list ready for business_days or date_series. Up to 10 years per call. Future one-off holidays are only included once proclaimed.
Input schema: {"type":"object","properties":{"country":{"type":"string","enum":["US","GB-EAW"],"description":"US = United States federal; GB-EAW = England and Wales bank holidays"},"fromYear":{"type":"integer","minimum":2000,"maximum":2200},"toYear":{"description":"default fromYear","type":"integer","minimum":2000,"maximum":2200}},"required":["country","fromYear"],"additionalProperties":false}

### fx_rate
Converts between about 30 currencies (USD, EUR, GBP, JPY, CAD, AUD, CHF, CNY, INR, MXN, BRL, ...) at the European Central Bank's euro reference rates, published once per business day around 16:00 CET. Cross rates go through the euro exactly as the ECB publishes them. Give a date (last 90 days) or get the latest; a weekend or holiday uses the last published rates and says which. Returns each rate, its inverse, the converted amount (rounded to the currency scale you choose) and the rate date. Reference rates are for information: banks and card networks add their own margins.
Input schema: {"type":"object","properties":{"from":{"type":"string","pattern":"^[A-Za-z]{3}$","description":"ISO 4217 code, e.g. USD"},"to":{"minItems":1,"maxItems":30,"type":"array","items":{"type":"string","pattern":"^[A-Za-z]{3}$"},"description":"one or more target codes, e.g. [\"EUR\", \"GBP\"]"},"amount":{"description":"amount to convert as a decimal string, default \"1\"","type":"string","maxLength":40},"date":{"description":"rates for this date (within the last 90 days); default the latest","type":"string","pattern":"^\\d{4}-\\d{2}-\\d{2}$"},"scale":{"description":"decimal places of converted amounts, default 2","type":"integer","minimum":0,"maximum":8},"rounding":{"description":"default half_up","type":"string","enum":["down","up","floor","ceil","half_up","half_even","half_down"]}},"required":["from","to"],"additionalProperties":false}

### slippage_bounds
From a quoted amount in base units (integer string) and a slippage tolerance in basis points, computes the minimum you should accept (min_out, rounded down) or the maximum you should pay (max_in, rounded up). Integer arithmetic only; the rounding direction is stated.
Input schema: {"type":"object","properties":{"amount":{"type":"string"},"slippageBps":{"type":"integer","minimum":0,"maximum":10000},"direction":{"type":"string","enum":["min_out","max_in"]}},"required":["amount","slippageBps","direction"],"additionalProperties":false}

### swap_price_impact
For a Uniswap-V2-style pool, computes the exact output of a swap (the router's integer formula), the total cost versus the spot price including the fee, and the part due to trade size alone, in basis points. Inputs are reserves and amounts in base units. Does not know any pool: you supply the reserves.
Input schema: {"type":"object","properties":{"reserveIn":{"type":"string"},"reserveOut":{"type":"string"},"amountIn":{"type":"string"},"feeBps":{"type":"integer","minimum":0,"maximum":10000}},"required":["reserveIn","reserveOut","amountIn"],"additionalProperties":false}

### money_calc
Rounds, adds, subtracts, multiplies, or splits an amount of an ISO 4217 currency, using that currency's real number of minor units (JPY 0, USD 2, KWD 3). A currency not in the list is refused, never assumed to have 2 decimals. split divides a total into equal parts or by ratios so the parts always add back to the total exactly.
Input schema: {"type":"object","properties":{"currency":{"type":"string"},"op":{"type":"string","enum":["round","add","subtract","multiply","split"]},"amount":{"type":"string"},"other":{"type":"string"},"factor":{"type":"string"},"parts":{"type":"integer","minimum":-9007199254740991,"maximum":9007199254740991},"ratios":{"type":"array","items":{"type":"string"}},"rounding":{"description":"rounding rule; default half_up (ties away from zero). half_even is banker's rounding","type":"string","enum":["down","up","floor","ceil","half_up","half_even","half_down"]}},"required":["currency","op","amount"],"additionalProperties":false}

### interest_calc
Exact simple or compound interest for a principal, an annual percentage rate and a term in years, compounded annually, semiannually, quarterly, monthly, weekly or daily. The number of periods must be whole; fractional periods are refused, not approximated. Also returns the effective annual rate.
Input schema: {"type":"object","properties":{"principal":{"type":"string"},"ratePercent":{"type":"string"},"years":{"type":"string"},"compounding":{"type":"string","enum":["simple","annual","semiannual","quarterly","monthly","weekly","daily"]},"scale":{"type":"integer","minimum":0,"maximum":18},"rounding":{"description":"rounding rule; default half_up (ties away from zero). half_even is banker's rounding","type":"string","enum":["down","up","floor","ceil","half_up","half_even","half_down"]}},"required":["principal","ratePercent","years","compounding"],"additionalProperties":false}

### loan_calc
Monthly payment, total interest and total paid for a fixed-rate amortizing loan, in exact arithmetic. The payment is rounded once to the currency scale and the final payment absorbs the rounding so the balance ends at exactly zero.
Input schema: {"type":"object","properties":{"principal":{"type":"string"},"annualRatePercent":{"type":"string"},"termMonths":{"type":"integer","minimum":1,"maximum":600},"scale":{"type":"integer","minimum":0,"maximum":8},"rounding":{"description":"rounding rule; default half_up (ties away from zero). half_even is banker's rounding","type":"string","enum":["down","up","floor","ceil","half_up","half_even","half_down"]}},"required":["principal","annualRatePercent","termMonths"],"additionalProperties":false}

### unit_convert
Converts length, mass, volume, time, data, energy, area, speed and temperature using exact definitions (an inch is exactly 0.0254 m). Ambiguous names (oz, gal, pt, KB) return AMBIGUOUS_INPUT with the choices instead of picking one. Reports whether the result is exact or rounded.
Input schema: {"type":"object","properties":{"value":{"type":"string"},"from":{"type":"string"},"to":{"type":"string"},"scale":{"type":"integer","minimum":0,"maximum":30},"rounding":{"description":"rounding rule; default half_up (ties away from zero). half_even is banker's rounding","type":"string","enum":["down","up","floor","ceil","half_up","half_even","half_down"]}},"required":["value","from","to"],"additionalProperties":false}

### address_format_check
Checks whether a string is a well-formed address for EVM (EIP-55 checksum), Solana, Aptos (AIP-40) or Bitcoin (base58check, bech32, bech32m). With family "auto" it lists every family the string could be, and never picks one. A valid format is not evidence an address is safe, owned, or the one you meant.
Input schema: {"type":"object","properties":{"address":{"type":"string"},"family":{"type":"string","enum":["evm","solana","aptos","bitcoin","auto"]}},"required":["address","family"],"additionalProperties":false}

### time_convert
Converts an ISO 8601 instant, or a local time plus its IANA zone, to another IANA zone. A local time skipped by a daylight-saving jump is refused; one that occurs twice returns AMBIGUOUS_INPUT with both instants. Zone abbreviations such as EST are refused.
Input schema: {"type":"object","properties":{"time":{"type":"string"},"fromZone":{"type":"string"},"toZone":{"type":"string"}},"required":["time","toZone"],"additionalProperties":false}

### business_days
Counts business days between two dates, adds business days to a date, or tests whether a date is a business day. Weekends are configurable. NO holidays are assumed: pass the holidays you want observed. Dates are calendar dates (YYYY-MM-DD) with no time zone.
Input schema: {"type":"object","properties":{"op":{"type":"string","enum":["count","add","is_business_day"]},"from":{"type":"string"},"to":{"type":"string"},"days":{"type":"integer","minimum":-9007199254740991,"maximum":9007199254740991},"weekend":{"type":"string","enum":["sat_sun","fri_sat","sun_only","none"]},"holidays":{"maxItems":400,"type":"array","items":{"type":"string"}}},"required":["op","from"],"additionalProperties":false}

### evm_balance
Native and ERC-20 balances of an address, ALL read at one stated block (number and hash returned). Token decimals are read from the token contract, never assumed; an address with no code, or a contract that is not a standard ERC-20, is an error and is never reported as a zero balance. Amounts are exact strings.
Input schema: {"type":"object","properties":{"chain":{"type":"string","description":"EVM CAIP-2 chain id such as eip155:8453, or a bare chain id number as a string. Chain names are refused."},"address":{"type":"string","description":"0x + 40 hex digits (EIP-55 checksum enforced when mixed-case)"},"assets":{"description":"\"native\" and/or token contract addresses; default [\"native\"]","minItems":1,"maxItems":20,"type":"array","items":{"type":"string"}},"block":{"description":"\"latest\" (default), \"safe\", \"finalized\", or a decimal block number. Every value in the answer is read at this one block","type":"string"}},"required":["chain","address"],"additionalProperties":false}

### evm_nonce
The transaction count of an address at one stated block. Pending-pool transactions are not counted, and the answer says so.
Input schema: {"type":"object","properties":{"chain":{"type":"string","description":"EVM CAIP-2 chain id such as eip155:8453, or a bare chain id number as a string. Chain names are refused."},"address":{"type":"string","description":"0x + 40 hex digits (EIP-55 checksum enforced when mixed-case)"},"block":{"description":"\"latest\" (default), \"safe\", \"finalized\", or a decimal block number. Every value in the answer is read at this one block","type":"string"}},"required":["chain","address"],"additionalProperties":false}

### evm_block
Number, hash, timestamp, base fee (gwei), gas used/limit and transaction count of one block. Use block "finalized" for a block that can no longer reorganize.
Input schema: {"type":"object","properties":{"chain":{"type":"string","description":"EVM CAIP-2 chain id such as eip155:8453, or a bare chain id number as a string. Chain names are refused."},"block":{"description":"\"latest\" (default), \"safe\", \"finalized\", or a decimal block number. Every value in the answer is read at this one block","type":"string"}},"required":["chain"],"additionalProperties":false}

### evm_tx_receipt
Status (success / reverted / pending), block, gas used and the exact fee in wei and native units for a transaction hash. 'Not found' is reported as such and is not proof the transaction never existed.
Input schema: {"type":"object","properties":{"chain":{"type":"string","description":"EVM CAIP-2 chain id such as eip155:8453, or a bare chain id number as a string. Chain names are refused."},"hash":{"type":"string"}},"required":["chain","hash"],"additionalProperties":false}

### evm_token_metadata
Name, symbol, decimals and total supply of an ERC-20 read from its contract at one block. Handles tokens whose name/symbol are bytes32. If decimals() is missing, it is an error: decimals are never assumed.
Input schema: {"type":"object","properties":{"chain":{"type":"string","description":"EVM CAIP-2 chain id such as eip155:8453, or a bare chain id number as a string. Chain names are refused."},"token":{"type":"string","description":"0x + 40 hex digits (EIP-55 checksum enforced when mixed-case)"},"block":{"description":"\"latest\" (default), \"safe\", \"finalized\", or a decimal block number. Every value in the answer is read at this one block","type":"string"}},"required":["chain","token"],"additionalProperties":false}

### evm_allowance
How much a spender may take from an owner for a token, at one block, in exact base units and human units, flagging 'unlimited' approvals.
Input schema: {"type":"object","properties":{"chain":{"type":"string","description":"EVM CAIP-2 chain id such as eip155:8453, or a bare chain id number as a string. Chain names are refused."},"token":{"type":"string","description":"0x + 40 hex digits (EIP-55 checksum enforced when mixed-case)"},"owner":{"type":"string","description":"0x + 40 hex digits (EIP-55 checksum enforced when mixed-case)"},"spender":{"type":"string","description":"0x + 40 hex digits (EIP-55 checksum enforced when mixed-case)"},"block":{"description":"\"latest\" (default), \"safe\", \"finalized\", or a decimal block number. Every value in the answer is read at this one block","type":"string"}},"required":["chain","token","owner","spender"],"additionalProperties":false}

### evm_contract_type
Says what is at an address using on-chain evidence: empty account, EIP-7702 delegated account, EIP-1167 minimal proxy, EIP-1967 proxy (with implementation and admin), Safe wallet (with owners and threshold), or other contract. Each classification lists the evidence. It is not a safety verdict.
Input schema: {"type":"object","properties":{"chain":{"type":"string","description":"EVM CAIP-2 chain id such as eip155:8453, or a bare chain id number as a string. Chain names are refused."},"address":{"type":"string","description":"0x + 40 hex digits (EIP-55 checksum enforced when mixed-case)"},"block":{"description":"\"latest\" (default), \"safe\", \"finalized\", or a decimal block number. Every value in the answer is read at this one block","type":"string"}},"required":["chain","address"],"additionalProperties":false}

### evm_logs
Logs emitted by one contract in a block range of at most 2000 blocks, optionally filtered by topics. Refuses (rather than truncates) when more than 1000 logs match, so an answer is always complete.
Input schema: {"type":"object","properties":{"chain":{"type":"string","description":"EVM CAIP-2 chain id such as eip155:8453, or a bare chain id number as a string. Chain names are refused."},"address":{"type":"string","description":"0x + 40 hex digits (EIP-55 checksum enforced when mixed-case)"},"fromBlock":{"type":"string"},"toBlock":{"type":"string"},"topics":{"maxItems":4,"type":"array","items":{"type":["string","null"]}}},"required":["chain","address","fromBlock","toBlock"],"additionalProperties":false}

### evm_calldata_decode
Decodes transaction calldata against an ABI that YOU supply and returns the function, its signature and each argument. If the selector is not in your ABI it says so and lists the selectors it does have; it does not guess from a selector database. An ABI that does not belong to the target contract can decode to plausible nonsense.
Input schema: {"type":"object","properties":{"abi":{"minItems":1,"maxItems":500,"type":"array","items":{"anyOf":[{"type":"object","propertyNames":{"type":"string"},"additionalProperties":{}},{"type":"string","maxLength":1000}]},"description":"JSON ABI objects, or human-readable signatures such as \"function transfer(address to, uint256 amount)\""},"data":{"type":"string"}},"required":["abi","data"],"additionalProperties":false}

### evm_log_decode
Decodes one event log (its topics and data) against an ABI that YOU supply and returns the event name and arguments. If no event in your ABI matches topic0 it says so rather than guessing from a database of known events.
Input schema: {"type":"object","properties":{"abi":{"minItems":1,"maxItems":500,"type":"array","items":{"anyOf":[{"type":"object","propertyNames":{"type":"string"},"additionalProperties":{}},{"type":"string","maxLength":1000}]},"description":"JSON ABI objects, or human-readable signatures such as \"function transfer(address to, uint256 amount)\""},"topics":{"minItems":1,"maxItems":4,"type":"array","items":{"type":"string"}},"data":{"type":"string"}},"required":["abi","topics","data"],"additionalProperties":false}

### payment_verify
Given a transaction hash, lists the ERC-20 transfers in its receipt and, if you give expectations (payee, asset, amount, payer), says whether a successful transaction contains a matching transfer. Use it to check an x402 or other USDC payment. A match proves tokens moved; it does not prove what the payment was for.
Input schema: {"type":"object","properties":{"chain":{"type":"string","description":"EVM CAIP-2 chain id such as eip155:8453, or a bare chain id number as a string. Chain names are refused."},"hash":{"type":"string"},"payTo":{"type":"string","description":"0x + 40 hex digits (EIP-55 checksum enforced when mixed-case)"},"asset":{"type":"string","description":"0x + 40 hex digits (EIP-55 checksum enforced when mixed-case)"},"amount":{"description":"base units","type":"string","pattern":"^(0|[1-9][0-9]{0,77})$"},"payer":{"type":"string","description":"0x + 40 hex digits (EIP-55 checksum enforced when mixed-case)"}},"required":["chain","hash"],"additionalProperties":false}

### random_from_block
Deterministic, publicly recomputable random integers derived from a FINALIZED block hash and a salt. Refuses a block that is not yet final. Not secret and not for cases where the salt can be chosen after seeing the block.
Input schema: {"type":"object","properties":{"chain":{"type":"string","description":"EVM CAIP-2 chain id such as eip155:8453, or a bare chain id number as a string. Chain names are refused."},"block":{"type":"string"},"salt":{"type":"string","minLength":1,"maxLength":200},"count":{"type":"integer","minimum":1,"maximum":100},"max":{"type":"integer","minimum":2,"maximum":1000000}},"required":["chain","block","salt","count","max"],"additionalProperties":false}

### erc8004_agent
Reads an agent from the ERC-8004 identity registry on Base: its owner address and the registration URI it points to, at one stated block. The URI is reported, not fetched (fetching arbitrary URLs for you is not this tool's job). An id that was never minted is NOT_FOUND.
Input schema: {"type":"object","properties":{"agentId":{"type":"string","pattern":"^(0|[1-9][0-9]{0,30})$","description":"ERC-8004 agent id (decimal string)"},"block":{"type":"string"}},"required":["agentId"],"additionalProperties":false}

### erc8004_reputation
Summarises the feedback posted about an agent in the ERC-8004 reputation registry on Base, at one block: number of distinct clients, number of feedback entries, the registry's own summary value, and each client's latest entry (value, tags, revoked). It is EVIDENCE, not a verdict: anyone can post feedback from any address, so many clients is not proof of many people, and tag text is free text written by the client (shown as untrusted).
Input schema: {"type":"object","properties":{"agentId":{"type":"string","pattern":"^(0|[1-9][0-9]{0,30})$","description":"ERC-8004 agent id (decimal string)"},"block":{"type":"string"}},"required":["agentId"],"additionalProperties":false}

### evm_spend_report
Every transfer of one ERC-20 token out of (or into) a wallet across a block range of up to 20,000 blocks, with exact amounts, per-counterparty totals, per-UTC-day totals, and an accounting CSV. Reads public chain data only. The range is queried in full 2000-block pieces and the report refuses (rather than truncates) when more than 2000 transfers match.
Input schema: {"type":"object","properties":{"chain":{"type":"string"},"wallet":{"type":"string"},"token":{"type":"string"},"fromBlock":{"type":"string"},"toBlock":{"type":"string"},"direction":{"type":"string","enum":["out","in"]}},"required":["chain","wallet","token","fromBlock","toBlock"],"additionalProperties":false}

### x402_settlement_verify
For sellers taking x402 or other stablecoin payments: give the network, the settlement transaction (EVM hash or Solana signature) and your payout address, and optionally the asset, minimum amount (base units) and payer. Reads the chain itself and answers verdict "paid", "not_paid" (failed, or no matching transfer to you) or "unknown" (not found yet), with every token transfer it saw. Networks: eip155:84532, eip155:8453, eip155:1, eip155:137, eip155:42161, eip155:10, eip155:43114, eip155:1329, eip155:130, eip155:146, eip155:480, eip155:5042, eip155:999, solana:5eykt4UsFv8P8NJdTREpY1vzqKqZKvdp, solana:EtWTRABZaYq6iMfeYKouRu166VU2xqa1. A match proves tokens reached you; it does not prove what they were for.
Input schema: {"type":"object","properties":{"network":{"type":"string","description":"CAIP-2 id, one of: eip155:84532, eip155:8453, eip155:1, eip155:137, eip155:42161, eip155:10, eip155:43114, eip155:1329, eip155:130, eip155:146, eip155:480, eip155:5042, eip155:999, solana:5eykt4UsFv8P8NJdTREpY1vzqKqZKvdp, solana:EtWTRABZaYq6iMfeYKouRu166VU2xqa1"},"tx":{"type":"string","description":"EVM transaction hash (0x + 64 hex) or Solana transaction signature (base58)"},"payTo":{"type":"string","description":"your payout address (EVM 0x... or Solana base58)"},"asset":{"description":"token contract (EVM) or mint (Solana) you expect; omit to accept any token","type":"string"},"amount":{"description":"minimum amount in base units (USDC: 1 USD = 1000000)","type":"string","pattern":"^(0|[1-9][0-9]{0,77})$"},"payer":{"description":"the address you expect the payment from","type":"string"}},"required":["network","tx","payTo"],"additionalProperties":false}

### token_price
Spot USD price of a crypto asset by ticker (BTC, ETH, SOL, USDC, APT, LINK, ...), read live from Coinbase and Kraken. A price is returned only when both answer and agree within 1%; the answer is their midpoint, with both readings and the deviation. If one venue is down or they disagree, you get an error with the readings instead of a guess. Ticker prices are for the market asset, not a specific token contract.
Input schema: {"type":"object","properties":{"symbol":{"type":"string","minLength":2,"maxLength":10,"description":"ticker such as BTC, ETH, SOL, USDC (case-insensitive)"}},"required":["symbol"],"additionalProperties":false}

### ens_resolve
Resolves an ENS name (e.g. vitalik.eth) to its Ethereum address, or an address to its primary ENS name, read live from Ethereum mainnet. Give exactly one of name or address. A reverse result is returned only if that name resolves back to the same address (otherwise it is reported as unverified, never as the answer). Names are normalized (ENSIP-15) first; a name that cannot be normalized is refused.
Input schema: {"type":"object","properties":{"name":{"type":"string","minLength":3,"maxLength":255},"address":{"type":"string"}},"additionalProperties":false}

### tx_safety_check
Before your agent signs an EVM transaction: simulates it on the current chain state and reports every token and native-coin movement in and out of the sender, decodes what it asks for (ERC-20 approve/permit, Permit2, setApprovalForAll, transfers), checks the counterparty (contract or wallet, sanctions lists) and, if you pass expectedTo, flags look-alike addresses (address poisoning). Verdict: "danger", "caution" or "no_known_risk", with every reason. It never says "safe": a simulation shows what would happen at this block. Chains: eip155:8453, eip155:1, eip155:137, eip155:42161, eip155:10, eip155:43114, eip155:1329, eip155:130, eip155:146, eip155:480, eip155:5042, eip155:999. Priced by the USD value at risk you declare in valueUsd: under $1,000: $0.005; under $10,000: $0.05; under $100,000: $0.50 (deep review); $100k and up: $2.00 (deep review). Deep review adds a second simulation on an independent provider and a counterparty check of every address involved.
Input schema: {"type":"object","properties":{"chain":{"type":"string","description":"CAIP-2 EVM chain id, e.g. eip155:8453"},"from":{"type":"string","description":"the address that would sign"},"to":{"type":"string","description":"the transaction's to address"},"data":{"description":"calldata, 0x... (omit for a plain native transfer)","type":"string"},"value":{"description":"native amount in wei (base units)","type":"string","pattern":"^(0|[1-9][0-9]{0,77})$"},"expectedTo":{"description":"the recipient or contract you MEANT to interact with; enables the look-alike check","type":"string"},"valueUsd":{"description":"approximate USD value at risk; sets the price tier, and $10,000+ includes the deep review","type":"number","minimum":0,"maximum":1000000000000}},"required":["chain","from","to"],"additionalProperties":false}

### sanctions_check
Checks whether an address appears on public sanctions lists: the US Treasury OFAC SDN digital-currency addresses (EVM, Bitcoin, Solana, Tron, Litecoin, Bitcoin Cash) and, for EVM addresses, Chainalysis' on-chain sanctions oracle. Returns sanctioned true/false, which sources matched, which were checked or unavailable, and when the list was last updated. This is a list match, not legal advice; no match is not a clearance. If no source can be read you get an error, never a 'clean' result.
Input schema: {"type":"object","properties":{"address":{"type":"string","minLength":20,"maxLength":120},"family":{"description":"auto-detected from the address format when omitted","type":"string","enum":["evm","solana","bitcoin","tron","litecoin","bitcoin_cash"]}},"required":["address"],"additionalProperties":false}

### counterparty_check
Before paying or delivering to an agent or address: checks sanctions lists, on-chain activity on Base and Ethereum (wallet, contract or EIP-7702 wallet; transaction counts), whether it owns an ERC-8004 agent identity, its record of deals verified by the LogicNodes notary (settled, and deliveries that failed their re-run check) and payments it made to LogicNodes. Returns every component, a 0-100 indicator and the formula: start 50; sanctions match = 0 (do not transact); +10 if 10+ transactions on Base or Ethereum, +10 more if 100+; +15 if it owns an ERC-8004 agent identity; +3 per deal settled through the LogicNodes notary (max +20); +1 per payment it made to LogicNodes (max +5); +10 if a domain operator is verified; -25 per delivery that failed its re-run check; clamped to 0-100. A new address has little history, which is not evidence of bad behaviour.
Input schema: {"type":"object","properties":{"address":{"type":"string","description":"EVM address of the agent or counterparty"}},"required":["address"],"additionalProperties":false}

### wallet_watch_create
Subscribes for 30 days: LogicNodes watches one EVM address and POSTs a signed webhook (X-LogicNodes-Signature: sha256=HMAC of the body with your secret) when tokens arrive (incoming) or leave (outgoing), when it grants a token or NFT approval (approval), or when its sanctions status changes (sanctions). Returns the watch id and a secret shown only once. Native-coin transfers are not watched. The webhook must be a public https URL. Chains: eip155:84532, eip155:8453, eip155:1, eip155:137, eip155:42161, eip155:10, eip155:43114, eip155:1329, eip155:130, eip155:146, eip155:480, eip155:5042, eip155:999.
Input schema: {"type":"object","properties":{"address":{"type":"string"},"network":{"type":"string","description":"CAIP-2 EVM chain id"},"webhookUrl":{"type":"string","maxLength":500,"description":"public https URL that receives the events"},"events":{"description":"default: all four","minItems":1,"maxItems":4,"type":"array","items":{"type":"string","enum":["incoming","outgoing","approval","sanctions"]}}},"required":["address","network","webhookUrl"],"additionalProperties":false}

### wallet_watch_status
Shows a watch's state (active, failing, cancelled, expired), the last block scanned, delivery counts and the last 20 events. Needs the watch id and its secret.
Input schema: {"type":"object","properties":{"watchId":{"type":"string","maxLength":80},"secret":{"type":"string","maxLength":200}},"required":["watchId","secret"],"additionalProperties":false}

### wallet_watch_cancel
Stops a watch immediately. Needs the watch id and its secret. (No refund: the subscription is pay-once.)
Input schema: {"type":"object","properties":{"watchId":{"type":"string","maxLength":80},"secret":{"type":"string","maxLength":200}},"required":["watchId","secret"],"additionalProperties":false}

### audit_log_create
Creates a logbook for an agent or business. Returns the log id, a write key (to append) and a read key (to export), shown only once. Every entry is hash-chained and timestamped into Bitcoin through the daily receipt root. Free to create; appending costs per entry.
Input schema: {"type":"object","properties":{"label":{"description":"a name for your own reference","type":"string","maxLength":120}},"additionalProperties":false}

### audit_log_append
Adds one entry (any JSON object up to 8 KB: a payment, a decision, a delivery, ...) to your log. Returns its sequence number, its chain hash and the receipt digest that will be in today's Bitcoin-timestamped root. Needs the log's write key.
Input schema: {"type":"object","properties":{"logId":{"type":"string","maxLength":60},"writeKey":{"type":"string","maxLength":100},"entry":{"type":"object","propertyNames":{"type":"string"},"additionalProperties":{}}},"required":["logId","writeKey","entry"],"additionalProperties":false}

### audit_log_export
Returns entries of your log (up to 1000 per call, from fromSeq), re-verifies the whole hash chain, and gives each entry's Bitcoin proof status (pending until its day's root is built and timestamped). Needs the log's read key.
Input schema: {"type":"object","properties":{"logId":{"type":"string","maxLength":60},"readKey":{"type":"string","maxLength":100},"fromSeq":{"type":"integer","minimum":1,"maximum":9007199254740991},"limit":{"type":"integer","minimum":1,"maximum":1000}},"required":["logId","readKey"],"additionalProperties":false}

### deal_dispute_review
For a deal recorded by the LogicNodes deal notary: verifies the signed, hash-chained record and gives an ADVISORY ruling from fixed rules (R0, R1, R2, R3, R4, R5, R6: settled, delivered and checked, failed check, uncheckable, never delivered, never agreed). Returns the facts it used (with event hashes), the rule applied, and a ruling digest that goes into the Bitcoin-timestamped daily root. Advisory only: not a legal judgment and not enforced.
Input schema: {"type":"object","properties":{"dealId":{"type":"string","pattern":"^0x[0-9a-fA-F]{64}$"}},"required":["dealId"],"additionalProperties":false}

### operator_verify_start
Step 1 (free) of the verified-operator badge. Returns one line to publish EITHER as the content of https://<domain>/.well-known/logicnodes-verify.txt OR as a DNS TXT record at _logicnodes.<domain>. Then call operator_verify_check. The badge proves control of the domain and is valid 365 days.
Input schema: {"type":"object","properties":{"address":{"type":"string","description":"the agent's EVM address"},"domain":{"type":"string","description":"e.g. example.com"}},"required":["address","domain"],"additionalProperties":false}

### operator_verify_check
Step 2 of the verified-operator badge: reads https://<domain>/.well-known/logicnodes-verify.txt and the DNS TXT record _logicnodes.<domain>; if either holds the line from operator_verify_start, the badge is recorded for 365 days and shows in counterparty_check and at /v1/operators/<address>. A failed check is an error (and errors are never charged).
Input schema: {"type":"object","properties":{"address":{"type":"string"},"domain":{"type":"string"}},"required":["address","domain"],"additionalProperties":false}

### agent_register_prepare
Step 1 (free): builds the ERC-8004 registration file and A2A agent card LogicNodes will host for your agent, and the exact message your agent address must sign (personal_sign) to publish them with agent_register_publish.
Input schema: {"type":"object","properties":{"address":{"type":"string","description":"the agent's EVM address (it signs the publish message and the registration transaction)"},"name":{"type":"string","minLength":2,"maxLength":80},"description":{"type":"string","minLength":10,"maxLength":1000},"url":{"description":"the agent's website","type":"string","maxLength":300,"format":"uri"},"mcpUrl":{"type":"string","maxLength":300,"format":"uri"},"a2aUrl":{"description":"the agent's own A2A JSON-RPC endpoint","type":"string","maxLength":300,"format":"uri"},"imageUrl":{"type":"string","maxLength":300,"format":"uri"},"x402":{"description":"true if the agent accepts x402 payments","type":"boolean"}},"required":["address","name","description"],"additionalProperties":false}

### agent_register_publish
Step 2: with the signature from agent_register_prepare, LogicNodes hosts the registration file at https://logicnodes.io/a/<address>/registration.json and the agent card next to it, and returns the unsigned transaction that calls register(tokenURI) on the ERC-8004 identity registry on Base (0x8004A169FB4a3325136EB29fA0ceB6D2e539a432). Your agent signs and sends it and pays the gas; LogicNodes never holds keys. Calling again with a new signature updates the hosted files.
Input schema: {"type":"object","properties":{"address":{"type":"string","description":"the agent's EVM address (it signs the publish message and the registration transaction)"},"name":{"type":"string","minLength":2,"maxLength":80},"description":{"type":"string","minLength":10,"maxLength":1000},"url":{"description":"the agent's website","type":"string","maxLength":300,"format":"uri"},"mcpUrl":{"type":"string","maxLength":300,"format":"uri"},"a2aUrl":{"description":"the agent's own A2A JSON-RPC endpoint","type":"string","maxLength":300,"format":"uri"},"imageUrl":{"type":"string","maxLength":300,"format":"uri"},"x402":{"description":"true if the agent accepts x402 payments","type":"boolean"},"signature":{"type":"string","pattern":"^0x[0-9a-fA-F]{130}$"}},"required":["address","name","description","signature"],"additionalProperties":false}

### x402_service_search
Searches every paid API listed in the public x402 directories (Coinbase CDP and PayAI, about 100,000 endpoints) and ranks the providers by what LogicNodes measures itself: whether the endpoint really answers with valid x402 payment terms (checked around the clock), uptime over 72 hours, USDC its payout address received on Base in the last 7 days and distinct payers. Returns matching endpoints with price, networks, method and an example input, so an agent can call the best one directly. Filter by network (CAIP-2), maximum price and live-only.
Input schema: {"type":"object","properties":{"query":{"type":"string","minLength":2,"maxLength":200,"description":"what you need, e.g. \"web scraping\", \"weather\", \"token price\""},"network":{"description":"only providers that accept payment on this CAIP-2 network, e.g. eip155:8453","type":"string","maxLength":80},"maxPriceUsd":{"description":"only providers with an endpoint at or below this price","type":"number","minimum":0,"maximum":1000},"liveOnly":{"description":"true: only providers whose last check returned valid x402 terms; false: everything; default: leave out providers whose last check failed (not-yet-checked ones are included, with checkedAt null, after the live ones)","type":"boolean"},"limit":{"description":"providers to return, default 5","type":"integer","minimum":1,"maximum":20}},"required":["query"],"additionalProperties":false}

### x402_provider_report
For one provider host (e.g. api.example.com): its listed paid endpoints and prices, the last checks (did it answer with valid x402 terms, how fast), 72-hour uptime, USDC received by its payout address on Base per day for 7 days, distinct payers, and its score with the formula. Use it before relying on a provider.
Input schema: {"type":"object","properties":{"host":{"type":"string","minLength":3,"maxLength":253,"description":"provider host, e.g. api.example.com (a full URL is accepted)"}},"required":["host"],"additionalProperties":false}

### x402_monitor_create
For sellers watching their own API, or buyers that depend on one: LogicNodes checks the provider every 5 minutes for 30 days and POSTs a signed webhook (X-LogicNodes-Signature: sha256=HMAC of the body with your secret) when it stops answering with valid x402 payment terms (two failed checks in a row) and when it recovers. Returns a monitor id and a secret shown once. Also shows on the provider's public page as monitored.
Input schema: {"type":"object","properties":{"host":{"type":"string","minLength":3,"maxLength":253,"description":"provider host from x402_service_search or the observatory"},"webhookUrl":{"type":"string","maxLength":500,"description":"public https URL that receives the alerts"}},"required":["host","webhookUrl"],"additionalProperties":false}

### x402_monitor_status
Shows a monitor's state, its provider's last checks and the last 20 alerts sent. Needs the monitor id and its secret.
Input schema: {"type":"object","properties":{"monitorId":{"type":"string","maxLength":80},"secret":{"type":"string","maxLength":200}},"required":["monitorId","secret"],"additionalProperties":false}

### x402_monitor_cancel
Stops a monitor immediately. Needs the monitor id and its secret. (No refund: the subscription is pay-once.)
Input schema: {"type":"object","properties":{"monitorId":{"type":"string","maxLength":80},"secret":{"type":"string","maxLength":200}},"required":["monitorId","secret"],"additionalProperties":false}

### x402_provider_claim
For sellers: prove you operate a provider by signing "LogicNodes observatory: I operate <host>" (personal_sign) with one of the EVM payout addresses its x402 listings use. The listing then shows as claimed on the public observatory page. Free.
Input schema: {"type":"object","properties":{"host":{"type":"string","minLength":3,"maxLength":253},"address":{"type":"string","description":"one of the provider's EVM payout addresses"},"signature":{"type":"string","pattern":"^0x[0-9a-fA-F]{130}$"}},"required":["host","address","signature"],"additionalProperties":false}

## Paid endpoints (x402)
- /premium/snapshot: $0.005 USDC. Gas price estimates for every supported chain in a single call.
- /premium/curve: $0.002 USDC. A dense gas-price curve (20 confidence points, 5-99%) for one chain, beyond the free tier's 12-level cap.
- /premium/forecast: $0.01 USDC. Deterministic worst/best-case base-fee bound for the next 1-10 blocks (EIP-1559's 12.5%-per-block cap), not a statistical prediction.
Pay in USDC on eip155:8453. An unpaid request returns 402 with a Payment-Required header.

## Quickstarts
### Gas prices (free, no key)
```bash
curl "https://logicnodes.io/gasprices/blockprices?chainid=8453"
```
```ts
const res = await fetch("https://logicnodes.io/gasprices/blockprices?chainid=8453");
const gas = await res.json();
console.log(gas.blockPrices[0].estimatedPrices[0]); // 99% confidence, gwei
```
```python
import requests
gas = requests.get("https://logicnodes.io/gasprices/blockprices", params={"chainid": 8453}, timeout=10).json()
print(gas["blockPrices"][0]["estimatedPrices"][0])  # 99% confidence, gwei
```

### Gas API v2 (multi-block, USD, history)
```bash
curl "https://logicnodes.io/v2/gas?chain=eip155:8453&confidenceLevels=99,90&blocks=3"
```
```ts
const r = await fetch("https://logicnodes.io/v2/gas?chain=eip155:8453&blocks=3");
const { ok, value, provenance, receipt } = await r.json();
console.log(value.estimates[0], value.usd.costs[0]);
```
```python
import requests
r = requests.get("https://logicnodes.io/v2/gas", params={"chain": "eip155:8453", "blocks": 3}, timeout=10).json()
print(r["value"]["estimates"][0], r["value"]["usd"]["status"])
```

### MCP: connect an agent
```bash
curl -s https://logicnodes.io/mcp -H "content-type: application/json" -H "accept: application/json, text/event-stream" \
  -d '{"jsonrpc":"2.0","id":1,"method":"tools/list"}'
```
```ts
// any MCP client; the server is stateless Streamable HTTP
{ "mcpServers": { "logicnodes": { "url": "https://logicnodes.io/mcp" } } }
```
```python
# MCP Python SDK (pip install mcp)
from mcp.client.streamable_http import streamablehttp_client
from mcp import ClientSession

async with streamablehttp_client("https://logicnodes.io/mcp") as (r, w, _):
    async with ClientSession(r, w) as s:
        await s.initialize()
        print(await s.list_tools())
```

### Multi-chain and Aptos tools (REST)
```bash
curl -s https://logicnodes.io/v1/tools/aptos_asset_resolve -H "content-type: application/json" -d '{"query":"USDC"}'
```
```ts
const r = await fetch("https://logicnodes.io/v1/tools/gas_estimate", {
  method: "POST", headers: { "content-type": "application/json" },
  body: JSON.stringify({ chain: "aptos:1" }),
});
const { ok, value, error, provenance, receipt } = await r.json();
```
```python
import requests
r = requests.post("https://logicnodes.io/v1/tools/gas_estimate", json={"chain": "aptos:1"}, timeout=15).json()
print(r["ok"], r.get("value") or r["error"])
```

### Paid endpoints with x402 (when payments are live)
```bash
# 1. unpaid call returns HTTP 402 and a base64 "Payment-Required" header describing price, asset, network and payTo
curl -i https://logicnodes.io/premium/snapshot
```
```ts
import { x402Client } from "@x402/core/client";
import { wrapFetchWithPayment } from "@x402/fetch";
import { registerExactEvmScheme } from "@x402/evm/exact/client";

const client = registerExactEvmScheme(new x402Client(), { signer: yourViemAccount });
const payFetch = wrapFetchWithPayment(fetch, client);
const res = await payFetch("https://logicnodes.io/premium/snapshot");
```
```python
# Plain HTTP flow any language can implement:
# 1) GET the URL -> 402 + Payment-Required header (base64 JSON: accepts[] with network, asset, amount, payTo)
# 2) sign a payment for one of the accepts[] entries with your wallet
# 3) retry the GET with the signed payload in the payment header; success returns 200 + Payment-Response
# See https://x402.org for client libraries.
```