Your agent shouldn't touch a Base token it hasn't checked.

🆕 Live — B20 is Base's native precompile token standard, and unlike ERC-20 a B20 issuer can freeze holders (Policy Registry) and even SEIZE your balance (burnBlocked) at the protocol level. This reads those powers — seizable, freezable, paused, rebase, uncapped-mint — into one hold/caution/avoid verdict, and separates a real tokenized stock from one merely wearing an equity ticker by reading the transfer-policy admin on chain, not a list: all 13 Coinbase equities, and the next on day one.

🆕 The single call before trading a Base-native B20: seize (burnBlocked) + freeze (Policy Registry) + rebase + pause + uncapped-mint, collapsed into one GO/HOLD/STOP verdict with an auditable receipt. Pass wallet= to also check if YOUR address is already blocked on that token. The B20 tool to bind first.

🆕 The per-transaction B20 rail check: pass token + from + to (+ optional executor) and get one GO/HOLD/STOP on whether this exact transfer clears NOW — sender policy resolved against the sender, receiver policy against the recipient, executor policy against the operator, plus live transfer-pause. Every other B20 tool is per-token due diligence bought once; this is the check an agent runs on every payment. No ERC-20 tool can see it.

🆕 Scans a wallet's B20 (Base-native) holdings for protocol-level freeze/seize powers and whether YOUR address is ALREADY blocked on any of them — the risk no ERC-20 portfolio tool can see. Returns per-token seizable/freezable/rebase flags plus a wallet-level verdict. Built for agents holding Base positions.

🆕 b20-safety tells you WHICH powers a B20 has; this tells you WHO holds them. Reads the token's role-based access control (mint / burn / seize-via-burnBlocked / pause / admin) from onchain role events and reports the exact controllers, how centralized they are, and whether admin has been renounced. The issuer-control map an agent needs before holding a regulated Base-native asset.

🆕 Every other B20 check reads what the issuer CAN do; this reads what they HAVE done. Scans the token's actual burnBlocked seizures (the BurnedBlocked event) — whether the issuer has ever burned a blocked holder's balance, who was seized, and how much. verdict: enforced (has seized) / armed (can, hasn't) / no_seize_power. Pass wallet= for one address, or omit address= for the network-wide feed.

🆕 Run FIRST on any 'B20'. A scammer can deploy a normal contract at a vanity 0xB200… address and fake the whole B20 read surface — the B20Factory precompile is the one authority that can't be spoofed. Verifies factory registration + bytecode absence (real B20s are chain-native precompiles with no code). verdict: genuine / fake_lookalike / not_b20. The 2-cent check that keeps every other B20 answer honest.

🆕 The B20 misconfiguration lint. Base's docs warn: a scope bound to a NON-EXISTENT allowlist silently denies EVERYONE — transfers brick. Audits every policy scope for dangling bindings, ALWAYS_BLOCK, renounced (frozen) lists and live pauses; verdict bricked / critical_misconfig / misconfigured / clean plus a can-this-token-even-move flag. Pre-launch lint for issuers, stuck-funds guard for holders and merchants.

🆕 b20-freeze-check answers one wallet; this enumerates the WHOLE list. Replays the Policy Registry's BlocklistUpdated/AllowlistUpdated events into the full membership of a B20's blocklist/allowlist: every address ever blocked or whitelisted, when, by whom, and the current member set per scope. Compliance-grade visibility no other tool provides. Pass address= (token) or policy= (registry policy ID).

🆕 createB20's initCalls run with role AND transfer-policy gates BYPASSED — a one-tx privilege window only the issuer ever gets. Reconstructs what they did in it: pre-mints (how much, to whom), role grants, policy bindings, blocklist seeding — the token's true starting conditions. verdict: blocklist_seeded / premined / configured_launch / bare_launch.

🆕 b20-supply shows the dilution ceiling; this shows the actual printing. Every mint (incl. batchMint) in the window: amount, recipients, and the share of current supply it represents — heavy_dilution / active_minting / minor_minting / quiet. The live issuance feed for agents holding B20 stablecoins and RWAs. Pass days= (1-90, default 30).

🆕 An Asset B20's multiplier rescales EVERY holder's balance in one call — no transfer, no per-account event, nothing in your tx history. b20-rebase shows today's value; this replays the full MultiplierUpdated history and flags DOWNWARD moves that silently cut every holder. The operator's rebase track record, before you hold a rebasing RWA.

🆕 A B20 Stablecoin's currency() is SELF-DECLARED — the standard verifies nothing. This is the missing check: declared peg vs actual DEX market price. 'Says USD, trades at $0.71' in one call: on_peg / depeg_warning / depegged / no_market (the peg is a pure claim with zero price discovery) / unverifiable_fx for non-USD pegs. Run before settling in any B20 stablecoin.

🆕 The premium tier of the B20 suite — no ERC-20 tool can produce it. Composes the full B20 picture: seize/freeze/pause/mint powers, WHO holds them (admin renounced?), allowlist-vs-blocklist access model, supply-cap dilution headroom, metadata mutability, and ACTUAL seizure history (burnBlocked) — then Claude writes an institutional due-diligence verdict: issuer-control score, seizure risk (enforced/armed/none), red flags, and a hold/avoid recommendation.

🆕 B20 adds memos to transfers/mints/burns (transferWithMemo) for payment IDs, compliance tags and settlement correlation — a field ERC-20 has no equivalent for. Reads a token's on-chain Memo event history, optionally filtered by a specific memo (bytes32) or the caller wallet. The settlement-reconciliation primitive for agents paying over B20 stablecoins.

🆕 The dilution half of B20 rug risk (b20-safety covers seizure). Reads supply cap vs minted supply (how much can still be minted) plus the on-chain SupplyCapUpdated history — an issuer that RAISED the cap diluted, or set up to dilute, holders. Uncapped mint is flagged as the worst case. Pair with b20-control to see who holds the mint role.

🆕 A B20 with a METADATA_ROLE holder can call updateName/updateSymbol — change its own identity after launch (an impersonation / bait-and-switch vector ERC-20 has no protocol equivalent for). Reads whether the metadata is mutable (role holder exists) AND whether it has ALREADY been renamed on-chain (NameUpdated/SymbolUpdated history). Trust the address, not the label.

🆕 Every B20 has ERC-2612 permit built in — approve a spender by signature instead of a transaction. Reads exactly what an agent needs to build a valid permit: the token's DOMAIN_SEPARATOR, the owner's current nonce (so the signed payload can't be rejected/replayed), and the EIP-712 domain/type struct. Read-only — signs nothing.

🆕 b20-control reads the token's own roles; the address that can actually add you to a blocklist lives in the Policy Registry. This reads WHO administers each active transfer/mint policy (policyAdmin), and whether that control is being handed over (pendingPolicyAdmin) or renounced. The other half of 'who can freeze/seize you', straight from the registry.

🆕 A B20 transfer policy is either a BLOCKLIST (allowed unless listed) or an ALLOWLIST (allowed ONLY if listed — a permissioned/whitelist token you can't even receive uninvited). b20-safety flags that a policy exists; this decodes its TYPE per scope (send/receive/execute/mint) — the difference between 'the issuer can block bad actors' and 'this is a permissioned RWA you can't hold uninvited'.

🆕 B20 Asset tokens can post on-chain announcements (Announcement: id, description, uri) — issuer notices, corporate actions, redemptions — a channel ERC-20 has no equivalent for. Reads a token's announcement feed (active vs ended) from CDP-indexed events. The issuer-communications primitive for agents holding tokenized/RWA B20 assets.

🆕 B20 Stablecoin tokens self-declare a fiat currency code (currency() → USD, EUR, …). Reads that declared peg alongside the issuance profile (supply, cap) and control powers (seize/freeze/pause/uncapped-mint) — a one-call 'what is this stablecoin and who controls it' for agents settling in B20 stablecoins. The currency code is the issuer's claim, not attested backing.

🆕 Lists the newest B20 tokens created on Base (from the B20Factory precompile) — variant, symbol, decimals, block. B20 is Base's native token standard and dozens launch hourly. A discovery feed for agents hunting new B20 launches early. Run b20-safety on any address before touching it — new ≠ safe.

The B20-only rug vector: a token can launch clean and later attach a sender blocklist (PolicyUpdated) — silently becoming seizable via burnBlocked. This reads the token's full policy/pause event timeline (CDP-indexed onchain events) plus the live policy state, and tells you if — and exactly WHEN — it turned seizable.

The live layer over B20 Policy Watch: a network-wide onchain webhook captures every B20 PolicyUpdated/Paused sub-second. Pass a token address for its live guard status, or call with no address for the feed of tokens that JUST attached a sender blocklist (turned seizable) across all of Base.

Complete B20 token profile straight from the precompile: variant (Asset/Stablecoin), name, symbol, decimals, total supply, supply cap, active transfer policies, pause states, and rebase. The data companion to b20-safety.

Checks whether a specific wallet is authorized under a B20 token's transfer-sender policy. If it isn't, that wallet can't transfer and can be burnBlocked() (SEIZED) by the issuer. The personal companion to b20-safety — 'can this token freeze MY funds?'

B20 Asset tokens do NOT apply multiplier() to balanceOf() — measured on chain: a multiplier moved 1.0→2.0 while holder balances read identically. Coinbase settles splits and dividend adjustments on its 13 tokenized equities through that multiplier, so reading balanceOf as a share count is wrong by exactly that factor the day one fires. Returns both numbers per holding: what an unadjusted integrator shows, and what is owed. Wallet balances only; LP/lending is declared excluded, never zeroed.

Reads a B20 Asset token's rebase multiplier — the factor that scales every holder's balance. Flags whether balances are being scaled (invisible dilution/inflation the issuer controls). Stablecoin variants return no-rebase.

Runs the B20 safety verdict across up to 5 B20 tokens in one call — each scored for freeze/seize/pause/rebase/uncapped-mint, with the worst score surfaced. For portfolio holders and agents screening several B20s at once.

Pass a URL and get the page as clean text — furniture (nav, scripts, ads, footers) stripped, paragraphs preserved — plus its title, description and site name. The step every agent needs before it can reason about a page, and the one most agents can't do themselves: serverless runtimes and MCP clients have no browser and no HTML parser. Follows redirects safely, refuses non-text content, caps at 2 MB. Nothing crypto about it.

One call replaces two: fetches the page, strips it to readable text, then returns exactly the fields you asked for as schema-enforced JSON (fields=price,author,published — up to 10). Add list=true to pull EVERY repeated record on the page (listings, table rows, search results) as an array — that mode costs $0.06, since it needs a much larger output budget. Built for agents that need a fact off a page, not a pile of HTML — no browser, no parser, no scraping code.

Search the live web from an agent that has no browser: pass query= and get back ranked results (title, URL, snippet, relevance score) plus a synthesized one-paragraph answer over them. maxResults=1..20 (default 5); includeAnswer=false to skip the summary and take sources only. The step before reasoning about anything that happened after the model's training cutoff — prices, releases, incidents, docs. Nothing crypto about it.

Exa's neural web search, resold per call — no Exa account, no subscription, one USDC micro-payment. Pass query= and get ranked results: title, URL, publish date, author, relevance score. numResults=1..10 (default 5). type=auto|fast|instant. includeDomains= restricts the search. Add text=1 for matched highlights from each page (caps results at 5, priced $0.03; the 402 challenge quotes it). For agents that need what the web says today, not what a training cutoff remembers.

Exa's own page index, read per call — no Exa account, no subscription. Pass urls= (one, or up to 5 comma-separated) and get each page as clean text with title, author and publish date. Answers from Exa's crawl rather than our fetcher, so it reads pages a plain server-side GET is blocked from. maxChars=500..40000 (default 8000). One URL is $0.002; a batch of 2-5 is $0.008, which the 402 challenge quotes. Pairs with exa-search: search, then read the winners.

Pass up to 5 URLs (comma-separated) and get each page back as clean text with its title. Renders pages that build their content in the browser, which a plain server-side fetch cannot read at all — so it covers the SPAs, dashboards and app pages that url-extract deliberately refuses. Returns partial success: pages that fail come back in a `failed` list with the reason, and a call where every URL fails is not charged. For a single static page, url-extract is cheaper.

Convert between the shapes data actually arrives in: CSV or TSV to JSON, JSON to CSV or TSV, CSV to a Markdown table, Markdown to HTML. Proper RFC 4180 parsing — quoted fields, embedded commas and newlines, doubled quotes — so a spreadsheet exported by a human does not silently come apart. Numbers and booleans get real types (turn it off with typed=false). No model call and no upstream: local parsing only.

Send Markdown (or HTML) and get back a URL that renders as a real page — headings, tables, code blocks, mobile-readable, no scripts and no external requests. The other half of an agent's output problem: file-slot is for bytes it will fetch back, this is for the report somebody has to read. Retention 1-30 days. Published on the storage origin, never on 402.com.tr.

Get a signed upload URL and a retrieval URL for one file, in one call. Your agent PUTs the bytes straight to storage — they never pass through this API, so there is no size ceiling imposed by a serverless runtime and no proxy in the middle. Declare bytes= and the size is signed into the URL. Up to 25 MB, retention 1-30 days, unguessable key. The step every agent hits the moment it produces a report, chart, CSV or build and has nowhere to put it.

Sanctions screening for NAMES, not wallets. Pass a person or company; get GO / HOLD / STOP against the official U.S. Treasury OFAC SDN export, including a.k.a. aliases. Conservative by design: only a full-name match is a hit. Screening aid, not a compliance determination.

Syntax + live DNS (MX/A) + disposable, role and consumer-mailbox detection, as one GO / HOLD / STOP verdict with the reasons listed. Catches throwaway inboxes, support@-style shared mailboxes and one-character domain typos (gmial.com). No SMTP probe, so mailbox existence is not claimed.

Registration age, expiry and registry status for any domain, from RDAP — the registries' own protocol. Returns days since registration, days to expiry, registrar, EPP status codes, nameservers and DNSSEC as one GO / HOLD / STOP verdict. Domain age is the strongest cheap signal in invoice and vendor-impersonation fraud: the payment details that change always arrive from a domain registered weeks ago. Refuses rather than guessing when a registry publishes no RDAP.

One call, four public sources: how old the domain is (RDAP), whether mail to the contact will land (DNS), whether the name is on the OFAC list, and whether the company exists in the global LEI register. Returns a single GO / HOLD / STOP with every signal that produced it. Built for the moment an agent has to trust an invoice, a new vendor or a payment-detail change — the classic fraud pattern is a real-looking name on a domain registered weeks ago.

🆕 Prices the same fiat purchase across EVERY payment method a country allows, and ranks them by what the buyer actually receives. On $100 in the US the spread is real: card takes 2.44, bank transfer 0.50, a funded fiat wallet nothing. Returns each rail's fee, net amount, min/max limits, and the cheapest one. Live Coinbase pricing, not a rate card. Pass address= for a payable checkout link. Not financial advice.

🆕 The feasibility check before you ask a user for money: which fiat-in and cash-out rails Coinbase actually runs in a country, the currencies and min/max limits for each, and how many assets across which networks. Distinguishes a country with card rails from one where crypto can only be transferred in — a difference most integrations discover only after a user fails. 122 countries covered.

Convert between 30+ currencies at the European Central Bank's daily reference rate, for today or any past date. Returns the converted amount, the per-unit rate and the date the rate is actually from — the ECB publishes on business days only, so a weekend date answers with the previous business day and says so instead of pretending. Reference rates for accounting and reporting, not tradable quotes.

Validate an IBAN offline: country registry length, structure, and the ISO 7064 mod-97 checksum that catches what humans and spreadsheets actually get wrong — a transposed pair, a dropped character, an O typed for a zero. Returns the normalised and print-formatted forms, country, SEPA membership, and the bank identifier where the layout is known. Every rejection names WHICH check failed, so you know whether to fix the row or drop it. Structure only: a valid IBAN is not proof the account exists.

Turn the many ways a phone number gets typed into one E.164 string: strips spacing, brackets, the 00 international prefix and the national trunk zero people include out of habit. Returns country, calling code, national number, and whether the length matches that country's numbering plan. Line type (mobile / fixed) is reported ONLY for countries whose ranges are unambiguous and is null elsewhere with a reason — never guessed. Offline: it does not check that the number is assigned or reachable.

Validate an EU VAT number offline: country prefix, national format, and the published check-digit algorithm for the 8 countries that have one (BE, DE, FR, IT, LU, NL, PL, PT). The response says which of the two you got — checksumChecked distinguishes a verified number from a format-only pass, because they are not the same strength of answer. Registration is deliberately NOT claimed: only VIES can say whether a number is registered, and this endpoint does not call it.

Add business days to a date, or count them between two dates, for ~110 countries. Knows each country's weekend (Friday-Saturday where that applies, not just Sat-Sun) and its national public holidays, and names every day it skipped. Holidays observed only in some provinces are deliberately NOT treated as national closures — they are listed separately so you can apply them when your case is regional. Built for payment terms, SLAs and delivery dates.

Restate an amount from one year in another year's money, for any country in the World Bank consumer-price series (1960 onwards). Returns the adjusted amount, the cumulative inflation and both index values so the arithmetic is checkable. The series is annual and published with a lag, so a request for a year that does not exist yet answers with the newest one available and says which. For contract comparisons, historical pricing and long-running budgets.

Look a company up in GLEIF, the global Legal Entity Identifier register: LEI code, registered legal name, entity status, registration status, jurisdiction and address. The register's own name filter is a token search that returns thousands of unrelated rows, so matches are re-checked here and only an exact one — or one differing solely in legal form (Inc/Ltd/GmbH) — is reported as found; the rest come back as candidates.

Validate a BIC/SWIFT code offline (ISO 9362): 8 or 11 characters, bank/country/location[/branch] layout, head-office vs branch, and the test-BIC flag. Every rejection names which check failed. Structure only — does not prove the bank is live.

Offline Issuer Identification Number check: accepts 6–8 digits only (refuses anything that looks like a full PAN), returns Visa/Mastercard/Amex/Discover/JCB/UnionPay/Maestro when the prefix is known. Not a live BIN registry lookup.

Validate a SEPA Creditor Identifier offline: country, check digits (mod-97), business code and national id. Catches mistyped creditor ids before a mandate setup. Structure only — not a scheme-manager registration proof.

Offline tax/customs id formats: EORI, US EIN/SSN, UK NINO, AU ABN (checksum), TR VKN/TCKN (checksum), CA BN. kind= selects the scheme. A well-formed id is not proof of registration.

Structural address check: country, street, city, and postal pattern for common countries (TR, US, GB, DE, FR, NL, …). Returns a normalised single-line form and named issues. Not proof of deliverability.

Forward geocode an address (q=) or reverse geocode lat=+lon= using OpenStreetMap Nominatim. Returns up to 5 hits with display name and address parts. Community map data — not a postal authority.

Guess UPS / FedEx / USPS / DHL / DPD / TNT / TR carriers from tracking-number shape and return a carrier tracking URL. Live parcel status is NOT fetched — pattern detection only.

Map a short product description to likely HS chapter/heading hints (electronics, apparel, coffee, pharma, …). Keyword table only — not a binding customs classification; confirm before filing.

TLS probe for host=: subject, issuer, validFrom/validTo, daysRemaining, expired/expiringSoon, fingerprint. GO/HOLD/STOP from dates. Chain trust is not asserted — dates and identity fields are.

Have I Been Pwned Pwned Passwords via k-anonymity: only a 5-char SHA-1 prefix leaves the server. Returns pwned + occurrence count and a GO/HOLD/STOP. Password is not stored. Email breach lookup not offered.

Heuristic URL risk: scheme, IP hosts, deep subdomains, credential-themed labels, sensitive paths/params, optional brand= lookalike. Returns reasons + GO/HOLD/STOP. Not a malware or Safe Browsing scan.

Heuristic extraction from invoice/receipt text: totals, amounts, VAT hints, IBANs, dates, emails, invoice numbers. Not OCR — convert a scan to text first, then send it here.

Find 30-minute UTC slots where every zone is inside a local window (default 09–17) on a given date. Pass zones= IANA list. Does not know holidays or personal calendars.

List public holidays for country= + year= from Nager.Date (~110 countries). Marks national vs regional entries. Same source business-days uses — for agents that need the raw calendar.

Encode text= into a QR code as SVG (default) or format=dataurl PNG. Offline, no upstream. Built for agents putting payment links, IBANs or URLs onto a page.

Validate UPC/EAN/GTIN checksum offline, then look up the barcode on Open Food Facts when it is a food/consumer item. Non-food codes often checksum-ok with no product match.

Turn a USDC settlement into bookkeeping lines: amount, accounts, memo, optional Base tx= enrichment (status, timestamp, to). Returns journal JSON plus a two-line CSV. Not tax advice.

Parse a bank CSV (comma/semicolon/tab, quoted fields) and map common headers (date/amount/debit/credit/balance/description, TR aliases) into one schema. Up to 500 rows. Best-effort.

The premium flagship. One call runs the FULL battery — contract risk, holder concentration, liquidity depth, EXIT liquidity (can you actually sell), OFAC sanctions — and Claude synthesizes an institutional verdict: safety score, explicit buy/sell tradeability, liquidity & holder assessments, factors, risks, positives and a recommendation. The value is the orchestration + AI synthesis you can't get in one call anywhere else.

The flagship wallet report: aggregates net worth, age/activity and recent transactions, then Claude synthesizes a verdict (e.g. fresh/risky → established/active) with key observations. One call, agent-ready wallet intelligence.

The flagship report: aggregates token risk, holder concentration, price/liquidity and OFAC sanctions, then Claude synthesizes a structured verdict (avoid → favorable) with key risks and positives. One call, agent-ready intelligence you can't get free.

The zoom-out flagship: aggregates trending and newly-listed Base tokens, then Claude writes a concise market brief — mood, highlights, new & notable launches, and cautions (fresh/unknown tokens flagged for rug risk). One call gives a trading agent market context instead of dozens of lookups.

Pulls a wallet's active token approvals and Claude writes a security audit: overall risk level, USD at risk, and concrete revoke recommendations (which token/spender and why). The wallet-safety check agents and users run before trusting an address.

Give a Base transaction hash and Claude explains what it actually did in plain English — the action, a risk read (failed tx, risky approval, high-value transfer), and notes. Turns raw calldata into an answer agents and humans can use.

Combines security flags with the verified ABI's function names, then Claude explains the contract's dangerous capabilities — owner can mint, pause, blacklist, upgrade or self-destruct — with a danger level. Goes beyond raw flags to read what the contract can actually do.

Compress up to 16K characters of anything — articles, transcripts, email threads, reports — into 3-5 precise bullet points, one micro-payment per call. No API key, no subscription, no prompt engineering: send text=, get bullets back as clean JSON. The digest step for agent pipelines that read more than they can carry in context.

Turn ANY text into clean, schema-enforced JSON: pass text= plus the fields you want (fields=name,email,price,date — up to 10) and get exactly those keys back, guaranteed-valid JSON via Claude structured outputs. Add list=true to extract EVERY repeated record (invoice lines, listings, table rows) as an array. Up to 16K chars per call. Not crypto-specific — the universal parse step for agent pipelines: pages, emails, receipts, logs. Many documents? ai-extract-batch: 10 in one call for $0.10.

The pipeline version of ai-extract: send up to 10 documents (text1=…&text2=… or texts= with ||| separators) and one field list — get one schema-enforced JSON object per document, in order, from a single paid call. 10 extractions for $0.10 instead of 10x the per-call overhead: built for agents parsing feeds, inboxes, scrape batches and receipt piles. Up to 6K chars per document.

Claude-quality translation as a pay-per-call primitive: send text= (up to 6K characters) and to= any language, get only the translation back — no notes, no wrapper prose, safe to pipe straight into the next step. Priced by length: $0.03 to 2,000 characters, $0.08 above — the 402 challenge quotes the price for the text you sent. No API key or subscription. Built for agents localizing content, parsing foreign-language sources, or serving multilingual users.

Pass a token contract address and get a risk score + flags (ERC-20 conformance, ownership renounce, upgradeable proxy) computed live from Base. Built for trading bots and agents that vet tokens before buying.

The single call an agent makes before touching a Base token: token risk + sellability (honeypot/tax) + route/price-impact + deployer reputation, collapsed into one GO/HOLD/STOP verdict with an auditable receipt. Cheaper than the four checks à la carte. This is the tool to bind first.

The check before the riskiest moment — signing. Decodes raw unsigned calldata (approve / permit / transfer / setApprovalForAll), shows the exact intent (who gets power over what, and whether it's UNLIMITED), and screens the destination + spender for OFAC sanctions and dangerous owner powers, in one GO/HOLD/STOP verdict + receipt. No simulation needed — pure decode + onchain risk. Catches the unlimited-approval drain vector before it's signed.

🆕 Base Account spend permissions let an app or agent spend a scoped, recurring allowance from a wallet — the primitive behind autonomous agent payments on Base. This reconstructs a wallet's ACTIVE spend permissions from onchain approve/revoke events and flags the dangerous ones: unlimited allowance, no expiry, unrecognized spender. The Base-native, agent-era drain check that ERC-20 approval tools can't see.

🆕 The complete fund-movement authority on a Base wallet in one call: ERC-20 approvals (a spender you approved can pull the token) PLUS Base Account spend permissions (the agent-era scoped recurring allowance) — one drain-surface verdict with the ERC-20 revoke queue to act on. Approval tools miss the spend permissions; spend-permission tools miss the approvals. The only combined check for agent wallets that must know their whole exposure before holding funds.

🆕 Since Pectra an EOA can carry EIP-7702 delegated code (the 0xef0100 designator): every call to the wallet executes the DELEGATE's contract with the wallet's funds — a malicious delegate is total takeover, invisible to approval tools. Reads the designator, resolves the delegate and verdicts it: not_delegated / delegated_known (Coinbase's EOA→SmartWallet path) / delegated_unknown (🚨) / smart_contract. The drain surface no other Base tool checks.

🆕 First tooling for Base's Commerce Payments protocol (AuthCaptureEscrow): reconciles two-phase auth/capture payments — in escrow, captured, charged, voided, refunded, or payer-RECLAIMABLE because the operator missed the capture window. Filter by payment= (infoHash), payer=, receiver= or operator=, or omit everything for the network feed. The order-reconciliation read merchants and agents need before trusting onchain commerce.

🆕 Commerce Payments flows are DRIVEN by an operator — it authorizes escrow, captures for merchants and takes the fees. This audits one before you rely on it: 90-day volumes, fees taken, capture-vs-reclaim record (missed capture windows = merchants left unpaid), distinct payers and merchants. verdict: healthy_operator / mixed / sloppy_operator / no_activity. No other tool reads this new Base commerce protocol.

🆕 Funding provenance in one call. Traces a Base wallet back to its EARLIEST transaction and resolves who first funded it: a recognized exchange/bridge (real on-ramp, lower risk), an anon EOA (possible sybil/burner — trace the cluster), or a contract. Returns the first funder, whether it's a contract or EOA, the initial value, and wallet age. The sybil/origin screen no other Base tool gives — the counterparty check before you transact. wallet= required. Not financial advice.

🆕 Cross-chain inflow detection. Reads a wallet's recent USDC MINTS and correlates them with Circle CCTP receives to tell you whether its USDC is freshly bridged in (and from which source chain — Ethereum, Arbitrum, OP, Polygon, Solana…) vs natively issued. Freshly bridged capital is new money / a possible cross-chain hop — a real signal for trading and liquidation agents that no other Base tool surfaces. wallet= required, days= optional (default 30, max 90). Not financial advice.

🆕 Before your agent sends, approves or signs to an address: one GO/REVIEW/STOP verdict. Screens OFAC sanctions, Coinbase/Basename identity, wallet vs contract, how long the address has existed, and whether contract logic can be swapped AFTER you send. Every factor returns what it found and what it could not read — a check that did not run never reads as clean, so GO means everything material was actually screened. Not financial advice.

🆕 The 'who am I dealing with?' check for agents. Reads Coinbase's ONCHAIN verification (an EAS attestation by verifications.coinbase.eth) — meaning the address is tied to a KYC'd Coinbase account, the strongest sybil-resistance signal on Base — plus the address's Basename. Returns a verified/named/anonymous verdict + trust score. Pair with sign-guard/spend-audit for a full pre-transaction gate. Identity is a signal, not proof of honesty.

🆕 Resolves the complete Base identity behind an address or Basename: resolved address, avatar, description, website and social handles (X/Twitter, GitHub, Farcaster, Discord) read from Base's L2 Resolver text records. Lets an agent turn a counterparty address into a real profile — beyond just the name. Complements address-trust (verification) with the human context.

Flow, not snapshot: the largest transfers of a Base token in the last 24h, classified by whether whales are sending INTO DEX pools (sell pressure) or pulling OUT (accumulation) — a net sell-pressure read that decays in hours. Complements holder-forensics ('who could dump') with 'is anyone dumping now'. CDP-indexed events.

Retention in one call: pass up to 10 tokens to snapshot them and get a watchId; call again with that watchId to get only the DELTAS since last time — liquidity ±%, price ±%, became-honeypot, sell-tax spiked, liquidity pulled. The second call is worth more than the first. Built for agents holding positions that re-check daily.

EOA vs contract, ETH + USDC balance, transaction count and activity level — straight from Base RPC. Useful for counterparty and wallet checks.

Returns current Base gas fees (baseFee, maxPriorityFee) and three tiered suggestions (slow, normal, fast) in Gwei, derived live from the Base RPC. Built for agents that time or cost-estimate transactions.

🆕 Beryl (2026-06-25) cut the single-proof withdrawal window from 7 days to 5, with a dual-proof fast path at ~1 day. Given a Base withdrawal-initiation tx, this decodes the L2ToL1MessagePasser event (the withdrawalHash, target & value needed to prove/finalize on L1) and estimates the finalization windows under the post-Beryl rules. For agents managing cross-chain liquidity.

Pass a Base token contract address and get the current USD price, 24h change, liquidity, and volume from DexScreener (highest-liquidity pair auto-selected). No API key required.

Provide a Base transaction hash and get a structured summary: from/to, ETH value, status, gas used, block, method selector, and nonce — straight from Base RPC.

The hard honeypot question, answered three ways: security simulation (honeypot, cannot-sell-all, sell/buy tax, transfer-pausable), a LIVE transfer simulation we run ourselves from a real holder (reverts/taxed?), and exit liquidity. Returns a hard canSell verdict with reasons. Selling is where rugs hide — check before you buy.

Goes past 'top-10 = X%' to classify the holder base: how much the creator & owner still hold, which top holders are infrastructure (LP/CEX/bridge — benign) vs unlabelled wallets (the concentration that can actually dump the price), and the largest non-infra wallet. Separating benign from dangerous concentration is the analysis others skip.

Reads the EIP-1967 proxy slots live: is the contract upgradeable, what's the current implementation, and WHO can upgrade it. Flags the dangerous case — an EOA admin that can swap the logic at any block with no timelock/multisig. Upgradeability is a rug vector static scans miss.

Reads a verified contract's ABI and flags owner-callable abuse functions: mint (dilute), pause (freeze exit), blacklist (block your wallet), setFee/setTax (tax you after entry), withdraw/sweep (pull funds), upgrade (swap logic). An unverified contract is itself a red flag. Diligence a price feed can't give.

Surfaces the LP holders, how much of the LP supply is locked or burned vs pullable, the lockers, and unlock dates. Unlocked liquidity is the clearest rug setup — this shows the detail a one-line 'LP locked: yes/no' hides.

Profiles the token's creator wallet: how much history it has, how much of the supply the creator still holds, and whether ownership is renounced — rolled into a 0-100 reputation score with signals. A fresh wallet holding 20% of supply with no renounce is the classic rug setup; this is the forensics layer other checks skip.

Turns raw LP-lock data into a forward calendar: each unlock with its date, the % of LP it frees, the locker (UNCX / Team Finance labelled), and days away — flagging imminent unlocks (<30 days). An LP unlock is a scheduled price event: the moment a rug becomes possible. For agents managing open positions.

The post-trade check everything else skips: given a token, your position size and (optionally) entry price, returns live price & P&L, whether the position can still be EXITED at that size, and the token's current rug score — combined into a healthy / watch / exit_now verdict with reasons. The risk that changes after you buy is exactly the risk holders miss.

Agents choose between tokens, not around them. Pass 2-5 Base token addresses → each gets a 0-100 quality score (safety-weighted: rug score first, then liquidity depth, then momentum), ranked best-first, with a named pick — or an honest 'none pass the gate'. One call returns a decision instead of a dozen data dumps.

Ranks a wallet's active token approvals by USD-at-risk x unlimited-allowance x unlabelled-spender, and returns a prioritised revoke queue. Approvals are the #1 drain vector; this tells an agent exactly what to revoke first.

Pulls a wallet's holdings and runs a risk check on each, flagging which positions are honeypots / high-risk / illiquid and the USD sitting in risky tokens. One call audits the whole wallet — 'which of the things you already hold could hurt you'.

Pay once to watch a Base token's liquidity. We snapshot a baseline and, on each monitor run, POST your webhook if liquidity collapses (a liquidity pull — the actual moment of a rug). Not price moving — the pool being drained out from under you. Expires after 30 days.

Returns native ETH plus WETH, USDC, USDbC, DAI, cbETH balances for any Base address, enriched with live USD values via DexScreener. Only non-zero holdings are returned.

Fetches the DexScreener boosts feed filtered to Base — up to 15 trending tokens with address, description, boost amount, and link. Great for discovery bots.

Pay once to register a price-threshold alert on any Base token. Supply target price, direction (above/below), and an https webhook URL. A polling cron (daily by default; configurable to ~minutes) checks DexScreener and POSTs your webhook when it crosses. Expires after 30 days. Or omit the webhook and poll with check=<id>: the verdict is THRESHOLD_MET, ok, FIRED, or unknown when the price feed could not be read.

Checks Sourcify for a Base contract's verification status and returns its ABI as function/event name lists + item count (full or partial match). No API key required.

Resolves a 4-byte function selector (e.g. 0x70a08231) to candidate human-readable signatures via 4byte.directory. Accepts a bare selector or full calldata.

Forward + reverse Basename resolution read from the Base L2 Resolver: turn jesse.base.eth into an address, or an address into its primary Basename. No API key required.

Forward + reverse ENS resolution on Ethereum mainnet: turn vitalik.eth into an address, or an address into its primary ENS name. Complements the Basename resolver for Base.

Computes the 4-byte selector for a function signature (e.g. transfer(address,uint256) → 0xa9059cbb). The inverse of the decoder — useful for agents building or matching calldata.

Checks an address against the OFAC SDN list of sanctioned digital-currency addresses. Built for compliance agents and bots that must screen counterparties before transacting. Direct-address match; list refreshed regularly. No API key required.

Top-10 holders with their %, holder count, concentration level (whale risk), and LP-lock %, from GoPlus. Lets agents judge how fairly a token is distributed before trading. No API key required.

Pass a comma-separated list of up to 10 Base token addresses and get USD price + 24h change for each in one call. Ideal for agents pricing a portfolio or watchlist.

Combined counterparty screening for an address: direct OFAC sanctions match, EOA/contract profile, and (for contracts) risk flags — rolled into a single recommendation (blocked / review / clear). Built for compliance agents.

Pass a comma-separated list of up to 25 addresses and get an OFAC sanctions result for each, plus the flagged subset. Built for compliance agents vetting whole counterparty lists in one call.

Deterministic 0-100 rug-probability score combining security flags, holder concentration, LP lock and liquidity depth — with the exact signals that drove it. A fast numeric gate for trading agents. Higher = riskier.

Screen up to 10 Base tokens in a single paid call — each gets a 0-100 rug-probability score, risk level and top signals, sorted riskiest-first. Built for agents triaging a watchlist or portfolio without paying per token.

Live Base chain snapshot: latest block, base fee + priority fee (Gwei), current ETH price, and the estimated USD cost of a simple ETH transfer. For agents timing or budgeting transactions.

Returns the latest tokens profiled on Base from the DexScreener feed — address, description, and links. Great for discovery bots hunting new launches early.

Complete ERC-20 portfolio for a Base address — every non-zero token with balance, metadata and live USD value, plus a total. Token discovery via CDP Data API (Coinbase) with Alchemy fallback; USD from DexScreener.

Complete token portfolio for a Base address with USD values: every ERC-20 the wallet holds, priced from live DEX liquidity, plus native ETH — sorted by value with a total. Holdings nobody quotes are listed and counted separately rather than silently valued at zero.

Wallet age in days, first & last activity timestamps, and outgoing transaction count for any Base address — read straight from the Base archive, not a 30-day feed, so it dates wallets of any age. Counts and last-activity cover what the wallet SENT; inbound transfers don't raise the nonce and aren't counted. Built for sybil screening and counterparty trust checks.

The latest transactions for a Base address — hash, time, from/to, ETH value, success — via Covalent. For agents tracking what a wallet is doing.

Lists a wallet's active token approvals — which contracts can spend its tokens, the allowance, USD value at risk and a risk factor (revoke.cash-style). Essential wallet-security check for agents and users.

Recent transfers of a specific token for a wallet — direction (in/out), amount, USD value, counterparty, tx hash and time, via Covalent. For agents tracking token flows and cost basis.

Lists the NFT collections a Base wallet holds — name, count and floor value — spam-filtered, via Covalent. For agents profiling a wallet's NFT exposure.

🆕 One block on Base — height, hash, parent, timestamp and age in seconds, transaction count, gas used against the limit, and the base fee in gwei. Takes a block number, a 0x… block hash, or latest / finalized / safe / earliest. A block that isn't mined yet answers found:false with the reason rather than an error. Raw chain data, no interpretation.

🆕 A Base transaction by hash, unedited: from, to, value in wei and ETH, nonce, gas, fee caps, type, and the 4-byte method selector with the calldata size. Flags a contract creation (null to) and a still-pending transaction (null block) rather than leaving you to infer either. Unknown hashes answer found:false, not an error.

🆕 The receipt for a Base transaction: success or reverted, block, gas used, effective gas price, the fee in wei and ETH, how many logs it emitted and which contracts emitted them. A contract deployment reports the new address. No receipt yet means pending, dropped, or never here — and it says which is possible instead of failing.

🆕 The next nonce for any Base address, plus its ETH balance and whether it is a contract. The count means different things either side of that line — transactions sent for a wallet, contracts deployed for a contract — so both are reported under their own name instead of one ambiguous number. What you need before building a transaction, or to tell a fresh address from a used one.

🆕 An ERC-20 balance on Base, in raw units and in decimals, with the token's symbol and decimals resolved in the same call. Zero is flagged explicitly, since it is the commonest answer and the one most easily confused with a failed read. An address that doesn't implement balanceOf answers found:false and says so.

🆕 totalSupply for any Base token, returned raw and scaled by its decimals, alongside the name and symbol. When a contract doesn't declare decimals the answer assumes 18 and marks that it assumed — a silent guess here quietly moves every downstream number by orders of magnitude. Non-token addresses answer found:false.

🆕 What sits at a Base address: contract or plain wallet, and if a contract, whether it declares ERC-721 or ERC-1155 via ERC-165 or carries the ERC-20 shape, plus name, symbol, decimals and code size. Interface support is read from the contract's own declaration first and only inferred when there is none — the ordering matters, because NFTs answer totalSupply() too.

🆕 A smart account's gas can be paid by a PAYMASTER instead of the account itself — invisible to every 'does this wallet spend ETH' heuristic. Reads a wallet's UserOperationEvents across BOTH EntryPoints (v0.6 + v0.7): is it a smart account, its op count and success rate, and WHO sponsors its gas (self vs which paymaster, per-sponsor share). A fully-sponsored account is typically app- or agent-operated. wallet= required, days= optional (default 30, max 90). Not financial advice.

🆕 Given a paymaster address, audits whether it's a healthy, active gas sponsor. Reads its UserOperationEvents across both EntryPoints: sponsored op volume, distinct accounts served, success rate, total gas sponsored, and concentration (the share coming from its single busiest app). The read to run before integrating a paymaster (Coinbase / Pimlico / Alchemy / custom), or judging who funds a counterparty's gas. paymaster= required, days= optional (default 30, max 90). Not financial advice.

🆕 Multisig / treasury intelligence. Given an address: is it a Gnosis Safe, its owners and M-of-N threshold, version, activity, and — critically — its enabled MODULES. A module can move the Safe's funds via execTransactionFromModule with no owner signatures, so each one is an address with unilateral control. Returns a multisig / single_signer / has_modules verdict: the counterparty check before you trust funds to a multisig. address= required. Not financial advice.

🆕 First lending tooling in the catalog. Reads a Morpho Blue position on Base (the largest lending venue on Base) and returns its liquidation health in one call: health factor, current vs liquidation LTV, and the exact collateral price drop that triggers liquidation. Built for borrowing agents and treasuries that need to know how close they are before the market cuts them — Morpho's own API is deferred. Pass wallet= and optionally market= (defaults to cbBTC/USDC). Not financial advice.

🆕 A MetaMorpho vault spreads your deposit across Morpho Blue markets — your real risk is WHERE the curator put it and WHO can move it. Reads the vault's live allocation (concentration, per-market collateral + liquidation LTV), idle share, performance fee, timelock, and control (curator/owner/guardian, and whether one address holds both). Returns a diversified / concentrated / control_risk verdict: the read before parking funds in a yield vault. vault= required. Not financial advice.

🆕 Built for liquidator / MEV searchers. Reconstructs the active borrower set on a Morpho Blue market from Borrow events, prices every position onchain in one multicall, and ranks them by liquidation health — flagging positions liquidatable NOW (health <= 1.0) and those one small move away, with the collateral price drop that tips each over. market= optional (defaults cbBTC/USDC), maxHealth= cutoff (default 1.1). Not financial advice.

Give a Base token + a trade size in USD → estimated buy AND sell price impact, whether you can unwind that size without collapsing the pool, and the largest safe exit. The hidden form of a rug isn't 'you can't buy' — it's 'you can't sell'. Built for trading agents sizing positions.

Give a token and a trade size in USD → the deepest Base pool to route through, estimated price impact, a suggested slippage tolerance + minimum-out, all gated on a honeypot/sell-tax check of the token you'd receive (no point routing into something you can't sell). Moves an agent from analysis to action in one call.

Reads the deepest pool's 24h volume, buy/sell counts, liquidity and price move, and scores how organic the activity looks. Volume 10x+ the pool's liquidity, near-perfect buy/sell symmetry, or big volume that moves the price nowhere are the classic wash-trading signatures used to bait buyers. Returns a 0-100 suspicion score and verdict.

Lists every DEX pool for a Base token (pair address, DEX, quote symbol, price, liquidity, 24h volume), sorted by liquidity. Tells agents where — and how deep — a token can be traded.

Given a Base DEX pair (pool) address, returns price, liquidity, 24h volume, buy/sell transaction counts and FDV from DexScreener. For agents analysing a specific pool's depth and activity.

Price change and trading volume across 1h, 6h and 24h windows for a Base token, plus a trend read (strong_up → strong_down). Lets agents gauge momentum, not just a single 24h number.

Metadata for a Base token: name, symbol, logo image, official website and social links (X/Telegram/etc.) plus price & liquidity — from DexScreener. For agents and UIs enriching a token.

Current floor price for a Base NFT collection (OpenSea / LooksRare) via Alchemy. Pass the collection contract address. Coverage is limited to marketplace-listed collections. For agents tracking NFT markets.

approval-advisor tells you WHAT to revoke; this builds the HOW: the ready-to-sign transaction (to + calldata) for approve(spender, 0) on a token, plus the live current allowance read from Base (flags unlimited, or 'already revoked'). Hand the result straight to a wallet or agent signer — the action half of approval hygiene.

Generate N cryptographically-strong, url-safe random tokens server-side — session IDs, API nonces, one-time codes, coupon secrets. Pass count= for how many. A tiny paid utility for agents that need entropy without a crypto library.

Buy a prepaid balance in a single x402 settlement and get a secret credit token. Send it as the `x-credit-token` header on any paid service and each call debits its price from your balance — no per-call signature, no settlement latency. Built for agents that fire many checks a minute. Tiers: $0.25 (starter), $1, $5 (+10%), $20 (+20%). The token is shown once; balance lasts 180 days.