Skip to content

QuickBuild token-saver tools + x402 MCP server

dev.workers.quynguyenbac.quickbuild/agent-toolsquickbuild.quynguyenbac.workers.devstreamable-httpv1.2.0Scanned

Free no-LLM token savers: URL→Markdown, outline, JSON trim, log compress, token count; x402 extras

40 tools: 2 do something that can't be undone, 33 change things and 5 only read.

Tools: 0 delete, 2 irreversible, 33 write, 5 read.
  • Reads private data · 0
  • Reads text others write · 37
  • Can send data out · 12
  • 8 findings, 1 high

Tools listed without sign-in. Remote posture

With Anlyon

2 tools here do something that can't be undone. Govern them.

  • Anlyon lists the tools and adopts none.
  • Each tool you adopt gets approvals and receipts.
  • Your agent never holds the key.
Govern this server →

A tool your agent calls directly is not routed through Anlyon and is not governed by it. Anlyon Radar only lists and labels, and these labels are the same either way.

What the QuickBuild token-saver tools + x402 MCP server can do

The QuickBuild token-saver tools + x402 MCP server is published by quickbuild.quynguyenbac.workers.dev, a domain the MCP Registry verified. It has 40 tools. It can do things that cannot be undone, such as sending, paying or publishing (translate and fx_rate) and create or change data (33 tools: launchcheck_full, website_audit, page_metadata, domain_intel and 29 more), but cannot delete data. 5 tools only read.

37 tools bring in text other people can write (launchcheck_full, website_audit, page_metadata and 34 more); 12 tools can send data out of your system (launchcheck_full, website_audit, page_metadata and 9 more).

The fixed checks matched 8 findings, 1 of them high severity (credentials-mention, model-directed-text, annotation-conflict and base64-blob). Each one comes with the text it matched.

Its remote endpoint lists its tools without sign-in, and its tools changed 3 times since Anlyon Radar first saw them.

Last scanned Nov 10, 2026, version 1.2.0.

Tools

Grouped by what each tool does, most consequential first. Open one for its reasons and full description; descriptions are the publisher's own words.

40 tools

Can't be undone 2Claude Code: ask
  • fx_rate[PAID $0.002 USDC per call via x402 on Base; errors are free] Foreign-exchange rates from one base currency (`from`, default USD) to a list of currencies (`to`, comma-separated, or…Reads text others writeCan send data out1 finding

    Why it is in this group

    • The input addresses people (input.to): what it sends cannot be taken back.
    • The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.

    What it can reach

    • untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
    • outboundThe input takes a destination (input.to): what the tool writes goes there.

    Findings

    • Warn

      The server marks "fx_rate" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".

      Matched text

      annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":true,"openWorldHint":true}; The input addresses people (input.to): what it sends cannot be taken back.

      What this check looks for
    Description, as the server wrote it
    [PAID $0.002 USDC per call via x402 on Base; errors are free] Foreign-exchange rates from one base currency (`from`, default USD) to a list of currencies (`to`, comma-separated, or "all"; default 14 majors incl. EUR, GBP, JPY, CNY, INR, KRW, VND), with `amount` converted. Live mode uses Coinbase public exchange rates (cached 15 s, updated about every minute, 160+ ISO 4217 currencies plus XAU/XAG), falling back to exchangerate-api daily rates. With `date=YYYY-MM-DD` it returns the ECB reference rates for that day (Frankfurter, ~30 major currencies, back to 1999-01-04). All params optional: with none it returns USD to 14 major currencies. Invalid codes = 400, never charged.

    Annotations the server declares

    • title: Fiat FX rates and conversion
    • readOnlyHint: true
    • destructiveHint: false
    • idempotentHint: true
    • openWorldHint: true
  • translate[PAID $0.005 USDC per call via x402 on Base; errors are free] LLM translation (Meta Llama 4 Scout on Workers AI, M2M100 fallback).Reads text others writeCan send data out1 finding

    Why it is in this group

    • The input addresses people (input.to): what it sends cannot be taken back.
    • The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.

    What it can reach

    • untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
    • outboundThe input takes a destination (input.to): what the tool writes goes there.
    • outboundThe input takes a URL (input.url): a request to a URL can carry data to whoever runs it.

    Findings

    • Warn

      The server marks "translate" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".

      Matched text

      annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":true,"openWorldHint":true}; The input addresses people (input.to): what it sends cannot be taken back.

      What this check looks for
    Description, as the server wrote it
    [PAID $0.005 USDC per call via x402 on Base; errors are free] LLM translation (Meta Llama 4 Scout on Workers AI, M2M100 fallback). Give `text`, or a `url` whose main content is extracted as Markdown and translated in blocks, preserving Markdown formatting, URLs and code blocks. Source language auto-detected (or set `from`). Up to 20,000 chars per call (default 8,000). All params optional: with none it translates a short Spanish sample to English. Languages: en,es,fr,de,it,pt,nl,ru,uk,pl,cs,ro,hu,el,bg,sv,da,no,fi,tr,ar,he,fa,hi,bn,ur,ta,zh,ja,ko,vi,th,id,ms,tl,sw.

    Annotations the server declares

    • title: Translate text or web page
    • readOnlyHint: true
    • destructiveHint: false
    • idempotentHint: true
    • openWorldHint: true
Changes things 33Claude Code: ask
  • address_brief[PAID $0.02 USDC per call via x402 on Base; errors are free] Rules-only (no LLM) merge of /api/base/token, /api/base/risk, /api/base/activity or /api/sol/wallet, /api/sol/token log…Reads text others writeNo findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    • The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.

    What it can reach

    • untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
    Description, as the server wrote it
    [PAID $0.02 USDC per call via x402 on Base; errors are free] Rules-only (no LLM) merge of /api/base/token, /api/base/risk, /api/base/activity or /api/sol/wallet, /api/sol/token logic. signals are plain reasons (owner not renounced, new contract, concentrated holders, mint authority present, fresh address...). Every response has disclaimer: "informational only, not investment advice"; unavailable values are null (never invented). Invalid input = 400, never charged.

    Annotations the server declares

    • readOnlyHint: true
    • destructiveHint: false
    • idempotentHint: true
    • openWorldHint: true
  • base_activity[PAID $0.003 USDC per call via x402 on Base; errors are free] One-call wallet profile: account type (EOA / EIP-7702 / contract), nonce, ETH + main token balances with USD (public B…Reads text others writeNo findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    • The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.

    What it can reach

    • untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
    Description, as the server wrote it
    [PAID $0.003 USDC per call via x402 on Base; errors are free] One-call wallet profile: account type (EOA / EIP-7702 / contract), nonce, ETH + main token balances with USD (public Base RPC), total transactions, token transfers, last 7/30 day activity and last seen, token count (free Blockscout API, null when unavailable). firstSeen is null (not cheaply available). Every response has disclaimer: "informational only, not investment advice"; values that cannot be obtained from free sources are null with a note (never invented). Invalid input = 400, never charged.

    Annotations the server declares

    • readOnlyHint: true
    • destructiveHint: false
    • idempotentHint: true
    • openWorldHint: true
  • base_block[PAID $0.001 USDC per call via x402 on Base; errors are free] Block header for Base mainnet (chain 8453) from public Base RPC: block number, hash, parent hash, timestamp, age in se…Reads text others writeNo findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    • The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.

    What it can reach

    • untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
    Description, as the server wrote it
    [PAID $0.001 USDC per call via x402 on Base; errors are free] Block header for Base mainnet (chain 8453) from public Base RPC: block number, hash, parent hash, timestamp, age in seconds, confirmations vs current head, transaction count, gas used/limit/utilization, base fee (gwei), fee recipient and size. `block` = latest (default), safe, finalized, a decimal or 0x-hex block number, or a 0x 32-byte block hash. Useful to check chain liveness/finality or timestamp an event before paying for /api/base/gas or /api/base/tx. Head tags cached 2 s, historical blocks 1 h. Invalid input = 400, unknown block = 404, never charged.

    Annotations the server declares

    • title: Base block lookup (latest/finalized/number/hash)
    • readOnlyHint: true
    • destructiveHint: false
    • idempotentHint: true
    • openWorldHint: true
  • base_gas_price[PAID $0.001 USDC per call via x402 on Base; errors are free] Live gas snapshot for Base mainnet (chain 8453) read from public Base RPC: latest block number/time/utilization, base …Reads text others writeNo findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    • The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.

    What it can reach

    • untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
    Description, as the server wrote it
    [PAID $0.001 USDC per call via x402 on Base; errors are free] Live gas snapshot for Base mainnet (chain 8453) read from public Base RPC: latest block number/time/utilization, base fee, eth_gasPrice, eth_maxPriorityFeePerGas, the last 10 blocks of base fees, slow/standard/fast EIP-1559 fee suggestions (10th/50th/90th percentile tips) in gwei and wei, and the estimated total cost (L2 execution + OP Stack L1 data-fee upper bound) of an ETH transfer, an ERC-20 transfer and a swap in ETH and USD (Chainlink ETH/USD on Base), plus a one-line `summary`, `baseFeeTrend` (rising/flat/falling vs 10-block average) and `congestion` (low/normal/high). No params. Cached 5 s, typically answers in well under 500 ms. Need ETH price + network health too? /api/base/snapshot ($0.003). Errors are never charged.

    Annotations the server declares

    • title: Base gas price and tx cost
    • readOnlyHint: true
    • destructiveHint: false
    • idempotentHint: false
    • openWorldHint: true
  • base_name[PAID $0.001 USDC per call via x402 on Base; errors are free] Name resolution from public RPC: *.base.eth via the Basenames registry on Base, *.eth via the ENS Universal Resolver o…Reads text others writeNo findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    • The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.

    What it can reach

    • untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
    Description, as the server wrote it
    [PAID $0.001 USDC per call via x402 on Base; errors are free] Name resolution from public RPC: *.base.eth via the Basenames registry on Base, *.eth via the ENS Universal Resolver on Ethereum mainnet; a 0x address returns its Basename primary name with forwardVerified. Every response has disclaimer: "informational only, not investment advice"; values that cannot be obtained from free sources are null with a note (never invented). Invalid input = 400, never charged.

    Annotations the server declares

    • readOnlyHint: true
    • destructiveHint: false
    • idempotentHint: true
    • openWorldHint: true
  • base_risk[PAID $0.005 USDC per call via x402 on Base; errors are free] Decision-support heuristic for a Base ERC-20: bytecode scan for admin functions (mint, pause, blacklist, setFee, upgra…Reads text others writeNo findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    • The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.

    What it can reach

    • untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
    Description, as the server wrote it
    [PAID $0.005 USDC per call via x402 on Base; errors are free] Decision-support heuristic for a Base ERC-20: bytecode scan for admin functions (mint, pause, blacklist, setFee, upgrade), owner() and paused() reads, proxy detection (public Base RPC), source verification, contract age and top-10 holder concentration (free Blockscout API, null when unavailable), DEX liquidity/volume (DexScreener). Returns riskScore, riskLevel and short reasons. Every response has disclaimer: "informational only, not investment advice"; values that cannot be obtained from free sources are null with a note (never invented). Invalid input = 400, never charged.

    Annotations the server declares

    • readOnlyHint: true
    • destructiveHint: false
    • idempotentHint: true
    • openWorldHint: true
  • base_token_batch[PAID $0.01 USDC per call via x402 on Base; errors are free] Batch version of /api/base/token for Base mainnet (chain 8453): pass up to 10 comma-separated contract addresses in `ad…Reads text others writeNo findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    • The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.

    What it can reach

    • untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
    Description, as the server wrote it
    [PAID $0.01 USDC per call via x402 on Base; errors are free] Batch version of /api/base/token for Base mainnet (chain 8453): pass up to 10 comma-separated contract addresses in `addresses` and get, per token, contract/EOA check, ERC-20 name/symbol/decimals/totalSupply, proxy implementation, owner() and renounced flag, well-known flag, and `market` from DexScreener (USD price, total liquidity, 24h volume/change, FDV, top pair; market=0 to skip). A bad token yields a per-item error instead of failing the call. Fixed $0.01 per call (vs $0.002 x 10). Metadata cached 1 h, market 60 s. Defaults to USDC, WETH, AERO. Invalid input = 400, never charged.

    Annotations the server declares

    • title: Base token batch (10 tokens + DEX price)
    • readOnlyHint: true
    • destructiveHint: false
    • idempotentHint: false
    • openWorldHint: true
  • base_token_info[PAID $0.002 USDC per call via x402 on Base; errors are free] Inspect a token or any address on Base mainnet: contract vs EOA (incl.Reads text others write1 finding

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    • The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.

    What it can reach

    • untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.

    Findings

    • Warn

      The input schema at properties.address.description contains a 42-character base64 string.

      Matched text

      inputSchema.properties.address.description: "...l: defaults to USDC 0x833589fCD6eDb6E08f4c7C32D4f71b54bdA02913."

      What this check looks for
    Description, as the server wrote it
    [PAID $0.002 USDC per call via x402 on Base; errors are free] Inspect a token or any address on Base mainnet: contract vs EOA (incl. EIP-7702 delegation), bytecode size, ERC-20 name/symbol/decimals/totalSupply (raw and formatted; bytes32 symbols handled), proxy implementation (EIP-1967, beacon, ZeppelinOS/FiatToken), owner() and whether it is renounced, whether it is a well-known Base token (guards against spoofed symbols), `market` (DexScreener: USD price from the deepest Base pair, total liquidity, 24h volume and change, FDV, top pair; cached 60 s; market=0 to skip), and optionally the balance of `holder`. Read live from public Base RPC; metadata cached 1 h. Many tokens? /api/base/token-batch (10 per call). All params optional: defaults to USDC on Base. Invalid input = 400, never charged.

    Annotations the server declares

    • title: Base ERC-20 token info
    • readOnlyHint: true
    • destructiveHint: false
    • idempotentHint: true
    • openWorldHint: true
  • base_token_verdict[PAID $0.01 USDC per call via x402 on Base; errors are free] Fixed documented point rules (returned in rules[]) over live public Base RPC reads, Blockscout and the free GeckoTermin…Reads text others writeNo findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    • The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.

    What it can reach

    • untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
    Description, as the server wrote it
    [PAID $0.01 USDC per call via x402 on Base; errors are free] Fixed documented point rules (returned in rules[]) over live public Base RPC reads, Blockscout and the free GeckoTerminal API (liquidity, volume, FDV, price change, top pool). Unavailable values are null and not scored (listed in unknown[]). No LLM. Every response has disclaimer: "informational only, not investment advice". Not a contract = 404, invalid = 400, never charged.

    Annotations the server declares

    • readOnlyHint: true
    • destructiveHint: false
    • idempotentHint: true
    • openWorldHint: true
  • base_wallet_balances[PAID $0.002 USDC per call via x402 on Base; errors are free] Balances of one address on Base mainnet in a single call: native ETH, the well-known tokens USDC, USDT, USDbC, DAI, WE…Reads text others write1 finding

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    • The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.

    What it can reach

    • untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.

    Findings

    • Warn

      The input schema at properties.address.description contains a 42-character base64 string.

      Matched text

      inputSchema.properties.address.description: "...tional: defaults to 0xd8dA6BF26964aF9D7eEd9e03E53415D37aA96045 (vitalik.eth)."

      What this check looks for
    Description, as the server wrote it
    [PAID $0.002 USDC per call via x402 on Base; errors are free] Balances of one address on Base mainnet in a single call: native ETH, the well-known tokens USDC, USDT, USDbC, DAI, WETH, cbETH, wstETH, AERO (non-zero only, or all with includeZero=true), plus up to 10 extra ERC-20 contracts you pass in `tokens` (symbol/name/decimals read on-chain). Also an always-present `usdc` block (balance + USD, also when 0), `txCount` (= nonce, transactions sent), account type (EOA, EIP-7702 delegated EOA, or contract) and a USD total for ETH/WETH (Chainlink) and stablecoins. Read live from public Base RPC via Multicall3; cached 10 s. All params optional: defaults to a well-known public address. Invalid input = 400, never charged.

    Annotations the server declares

    • title: Base wallet balances (ETH + tokens)
    • readOnlyHint: true
    • destructiveHint: false
    • idempotentHint: false
    • openWorldHint: true
  • crypto_price[PAID $0.001 USDC per call via x402 on Base; errors are free] Live crypto prices for up to 25 coins per call: pass tickers in `symbols` (BTC,ETH,SOL,AERO,...; common tickers built …Reads text others writeNo findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    • The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.

    What it can reach

    • untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
    Description, as the server wrote it
    [PAID $0.001 USDC per call via x402 on Base; errors are free] Live crypto prices for up to 25 coins per call: pass tickers in `symbols` (BTC,ETH,SOL,AERO,...; common tickers built in, others resolved by CoinGecko search to the largest coin with that ticker), CoinGecko `ids`, or one token `address` + `chain` (base default; ethereum, arbitrum, optimism, polygon, bsc, avalanche, solana). Returns price, market cap, 24h volume and 24h % change in each `vs` currency (usd default; up to 5, e.g. usd,eur,btc) plus last-updated time. Each USD price is cross-checked against Coinbase spot: `sources` {coingecko, coinbase} and `deviationPct` per coin, `maxDeviationPct` overall. Data: CoinGecko public API, cached 15 s; falls back to Coinbase spot rates (price only) if CoinGecko is unavailable. All params optional: with none it returns BTC and ETH in USD. Unknown coins = 404, invalid input = 400, never charged.

    Annotations the server declares

    • title: Crypto prices (ticker, id or contract)
    • readOnlyHint: true
    • destructiveHint: false
    • idempotentHint: true
    • openWorldHint: true
  • domain_intel[PAID $0.003 USDC per call via x402 on Base; errors are free] One call domain lookup for agents: DNS records A, AAAA, MX, NS, TXT, CAA (Cloudflare DoH), SPF parsed (all qualifier, …Reads text others writeNo findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    • The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.

    What it can reach

    • untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
    Description, as the server wrote it
    [PAID $0.003 USDC per call via x402 on Base; errors are free] One call domain lookup for agents: DNS records A, AAAA, MX, NS, TXT, CAA (Cloudflare DoH), SPF parsed (all qualifier, lookup count, issues), DMARC parsed (policy, pct, rua, issues), RDAP summary (registrar, created, expires, updated, status, DNSSEC via rdap.org) and a 0-100 email-security score with letter grade and reasons. Param name = public domain (example.com; URLs are reduced to their host). IPs/internal names refused (400), NXDOMAIN 404, DNS upstream down 502: never charged.

    Annotations the server declares

    • title: Domain intel: DNS, SPF/DMARC, RDAP, email-security grade
    • readOnlyHint: true
    • destructiveHint: false
    • idempotentHint: true
    • openWorldHint: true
  • estimate_tokens[FREE] Fast approximate token count of text (heuristic estimate, not a real tokenizer; no prices).No findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    • The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
    Description, as the server wrote it
    [FREE] Fast approximate token count of text (heuristic estimate, not a real tokenizer; no prices). Limit 120/hour per IP.

    Annotations the server declares

    • title: Estimate tokens (free)
    • readOnlyHint: true
    • destructiveHint: false
    • idempotentHint: true
    • openWorldHint: false
  • free_site_check[FREE, 10 calls/hour per IP] Quick check of one public website: homepage meta/OG/canonical, security headers, HTTPS, and up to 20 links tested for breakage.Reads text others writeCan send data outNo findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    • The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.

    What it can reach

    • untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
    • outboundThe input takes a URL (input.url): a request to a URL can carry data to whoever runs it.
    Description, as the server wrote it
    [FREE, 10 calls/hour per IP] Quick check of one public website: homepage meta/OG/canonical, security headers, HTTPS, and up to 20 links tested for breakage. For the deeper 3-page audit with leaked-key scan use website_audit (paid).

    Annotations the server declares

    • title: Free quick website check
    • readOnlyHint: true
    • destructiveHint: false
    • idempotentHint: true
    • openWorldHint: true
  • generate_image[PAID $0.01 USDC per call via x402 on Base; errors are free] Generate an original 1024x1024 image from a text prompt with FLUX.1 [schnell] on Workers AI (about 2-5 s).Reads text others write1 finding

    Why it is in this group

    • The name "generate_image" says "generate": a word for a change.
    • The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.

    What it can reach

    • untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.

    Findings

    • Warn

      The server marks "generate_image" read-only (readOnlyHint: true), and its own name, description or input give the class "write".

      Matched text

      annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":true}; The name "generate_image" says "generate": a word for a change.

      What this check looks for
    Description, as the server wrote it
    [PAID $0.01 USDC per call via x402 on Base; errors are free] Generate an original 1024x1024 image from a text prompt with FLUX.1 [schnell] on Workers AI (about 2-5 s). Returns JSON with image_base64 + mimeType, or the raw image bytes with format=jpeg. Content policy: no sexual/nude, gore, hate, real or famous people, or ID documents/currency; such prompts get HTTP 400 and are not charged. All params optional: with no prompt a sample watercolor lighthouse is drawn.

    Annotations the server declares

    • title: Generate image (FLUX.1 schnell)
    • readOnlyHint: true
    • destructiveHint: false
    • idempotentHint: false
    • openWorldHint: true
  • injection_check[PAID $0.003 USDC per call via x402 on Base; errors are free] For developers of AI agents that read untrusted web pages, emails or tool outputs.Reads text others writeCan send data out1 finding

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    • The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.

    What it can reach

    • untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
    • outboundThe input takes a URL (input.url): a request to a URL can carry data to whoever runs it.

    Findings

    • High

      The description contains "ignore previous instructions", an instruction to set earlier instructions aside.

      Matched text

      description: "...promise): flags instruction overrides ("ignore previous instructions"), role/system impersonation (<|im_star..."

      What this check looks for
    Description, as the server wrote it
    [PAID $0.003 USDC per call via x402 on Base; errors are free] For developers of AI agents that read untrusted web pages, emails or tool outputs. A RULE/PATTERN filter (regex heuristics, no LLM, no model-quality promise): flags instruction overrides ("ignore previous instructions"), role/system impersonation (<|im_start|>, [INST], "system:"), exfiltration requests (send keys/secrets to a URL), hidden text (zero-width/bidi/Unicode-tag chars, HTML comments, display:none / white text, base64 blobs that decode to instructions), tool-call / shell command injection, markdown image/link exfil, and overrides in EN/ES/FR/DE/ZH/JA/RU. POST JSON {"text"} (max 32 KB) or GET ?text= for short text, or {"url"} to fetch one public https page (port 443, public hosts only, max 2 redirects, 64 KB read, 4 s timeout). Returns verdict, score 0-100, findings (category, snippet <= 80 chars with secrets redacted, position) and an honest `limits` note. Paraphrased or novel attacks can be missed; not a guarantee. Missing/oversized input = 400/413, unreachable page = 502/504, never charged.

    Annotations the server declares

    • title: Prompt-injection check (rule/pattern filter, no LLM) for text or a page
    • readOnlyHint: true
    • destructiveHint: false
    • idempotentHint: true
    • openWorldHint: true
  • launchcheck_full[PAID $0.05 USDC per call via x402 on Base; errors are free] LaunchCheck full report for one public URL: status code, HTTPS/redirects, title/description, Open Graph/Twitter card, r…Reads text others writeCan send data outNo findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    • The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.

    What it can reach

    • untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
    • outboundThe input takes a URL (input.url): a request to a URL can carry data to whoever runs it.
    Description, as the server wrote it
    [PAID $0.05 USDC per call via x402 on Base; errors are free] LaunchCheck full report for one public URL: status code, HTTPS/redirects, title/description, Open Graph/Twitter card, robots.txt/sitemap/noindex, plus up to 50 links checked for breakage, og:image load check, desktop 1280x800 and mobile 390x844 JPEG screenshots, edge TTFB and real-browser load time, image alt text, heading outline, mobile overflow, and a P1-P3 prioritized fix list. Read-only. Unreachable sites are not charged.

    Annotations the server declares

    • title: LaunchCheck full pre-launch report
    • readOnlyHint: true
    • destructiveHint: false
    • idempotentHint: true
    • openWorldHint: true
  • llm_chat[PAID $0.003 USDC per call via x402 on Base; errors are free] One-shot LLM completion on Cloudflare Workers AI with a minimal interface: give `prompt` (and optionally `system`, or …Reads text others writeNo findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    • The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.

    What it can reach

    • untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
    Description, as the server wrote it
    [PAID $0.003 USDC per call via x402 on Base; errors are free] One-shot LLM completion on Cloudflare Workers AI with a minimal interface: give `prompt` (and optionally `system`, or OpenAI-style `messages`), get {text, model, finish_reason, usage}. Models: llama-3.1-8b (default), llama-3.2-3b, llama-3.2-1b, mistral-7b, granite-4.0-micro. max_tokens capped at 512, input up to 24000 chars, Llama Guard 3 safety check, 503 "capacity" when the daily budget is used up. Errors are never charged. All params optional: with none it answers a short sample question. For the OpenAI wire format use POST /v1/chat/completions.

    Annotations the server declares

    • title: LLM chat completion (Llama 3.1 8B default)
    • readOnlyHint: true
    • destructiveHint: false
    • idempotentHint: false
    • openWorldHint: true
  • page_metadata[PAID $0.005 USDC per call via x402 on Base; errors are free] Fetch one public web page and return structured metadata: title, meta description, canonical, lang, robots, Open Graph…Reads text others writeCan send data outNo findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    • The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.

    What it can reach

    • untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
    • outboundThe input takes a URL (input.url): a request to a URL can carry data to whoever runs it.
    Description, as the server wrote it
    [PAID $0.005 USDC per call via x402 on Base; errors are free] Fetch one public web page and return structured metadata: title, meta description, canonical, lang, robots, Open Graph and Twitter card tags, favicon, RSS/Atom feeds, JSON-LD @types, and internal/external links (up to 100 each). Good for link previews, unfurling and quick site triage.

    Annotations the server declares

    • title: Page metadata / link preview
    • readOnlyHint: true
    • destructiveHint: false
    • idempotentHint: true
    • openWorldHint: true
  • rerank_documents[PAID $0.002 USDC per call via x402 on Base; errors are free] Cohere/Jina-style rerank: POST {query, documents: string[] | {text}[], top_n, return_documents} and get {results: [{in…Reads text others writeNo findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    • The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.

    What it can reach

    • untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
    Description, as the server wrote it
    [PAID $0.002 USDC per call via x402 on Base; errors are free] Cohere/Jina-style rerank: POST {query, documents: string[] | {text}[], top_n, return_documents} and get {results: [{index, relevance_score, document?}]} sorted by relevance. Models (Cloudflare Workers AI): bge-reranker-base (cross-encoder, English/Chinese, scores 0..1); bge-m3 (multilingual similarity scoring) (default bge-reranker-base; other names map to it). Up to 100 documents, each scored on its first 4000 chars; one fixed price per call. Use after vector search to pick the best passages for an LLM. Errors are never charged. All params optional: with none it ranks a small sample.

    Annotations the server declares

    • title: Rerank documents for a query
    • readOnlyHint: true
    • destructiveHint: false
    • idempotentHint: true
    • openWorldHint: true
  • sol_token[PAID $0.003 USDC per call via x402 on Base; errors are free] Mint account read live from the public Solana mainnet RPC (jsonParsed).Reads text others writeNo findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    • The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.

    What it can reach

    • untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
    Description, as the server wrote it
    [PAID $0.003 USDC per call via x402 on Base; errors are free] Mint account read live from the public Solana mainnet RPC (jsonParsed). Every response has disclaimer: "informational only, not investment advice"; unavailable values are null (never invented). Not a mint = 404, invalid = 400, never charged.

    Annotations the server declares

    • readOnlyHint: true
    • destructiveHint: false
    • idempotentHint: true
    • openWorldHint: true
  • sol_wallet[PAID $0.003 USDC per call via x402 on Base; errors are free] Live from the public Solana mainnet RPC (no API key).Reads text others writeNo findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    • The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.

    What it can reach

    • untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
    Description, as the server wrote it
    [PAID $0.003 USDC per call via x402 on Base; errors are free] Live from the public Solana mainnet RPC (no API key). Tokens sorted known-first then by amount; USD values not computed. Every response has disclaimer: "informational only, not investment advice"; unavailable values are null (never invented). Invalid input = 400, never charged.

    Annotations the server declares

    • readOnlyHint: true
    • destructiveHint: false
    • idempotentHint: true
    • openWorldHint: true
  • text_embeddings[PAID $0.001 USDC per call via x402 on Base; errors are free] Drop-in OpenAI Embeddings endpoint: POST {input: string | string[], model, encoding_format: float|base64, dimensions} …Reads text others writeNo findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    • The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.

    What it can reach

    • untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
    Description, as the server wrote it
    [PAID $0.001 USDC per call via x402 on Base; errors are free] Drop-in OpenAI Embeddings endpoint: POST {input: string | string[], model, encoding_format: float|base64, dimensions} and get {object: "list", data: [{embedding, index}], model, usage}. Models (Cloudflare Workers AI): bge-m3 (1024-d, multilingual (100+ languages), 8192-token context); bge-large-en-v1.5 (1024-d, English); bge-base-en-v1.5 (768-d, English); bge-small-en-v1.5 (384-d, English, fastest); embeddinggemma-300m (768-d, multilingual (Google EmbeddingGemma)). Any other model name (e.g. text-embedding-3-small) maps to bge-m3; the actual model is in the response. `dimensions` truncates and re-normalizes. Up to 100 inputs per call, 8000 chars each, 60000 chars total, one fixed price per call. For RAG, semantic search, clustering and dedupe. Errors are never charged. GET ?input= also works; with no params it embeds a sample sentence.

    Annotations the server declares

    • title: Text embeddings (bge-m3, OpenAI format)
    • readOnlyHint: true
    • destructiveHint: false
    • idempotentHint: true
    • openWorldHint: true
  • text_to_speech[PAID $0.005 USDC per call via x402 on Base; errors are free] Turn text (up to 2000 chars) into spoken audio with MeloTTS on Cloudflare Workers AI, in English, Spanish, French, Chi…Reads text others writeNo findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    • The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.

    What it can reach

    • untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
    Description, as the server wrote it
    [PAID $0.005 USDC per call via x402 on Base; errors are free] Turn text (up to 2000 chars) into spoken audio with MeloTTS on Cloudflare Workers AI, in English, Spanish, French, Chinese, Japanese or Korean. Returns JSON with audio_base64 + mimeType (audio/wav, 44.1 kHz mono), or the raw audio bytes with format=wav. Good for voice replies, narration, accessibility and language practice. Text with sexual, hateful or violent content is refused (HTTP 400, not charged). All params optional: with none it speaks a short English sample. GET with query params or POST a JSON body.

    Annotations the server declares

    • title: Text to speech (WAV)
    • readOnlyHint: true
    • destructiveHint: false
    • idempotentHint: true
    • openWorldHint: true
  • trim_json[FREE] Shrink a JSON document for LLM context: drop null/empty values, truncate long arrays and strings (with counts), limit depth, optional key allowlist.No findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    • The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
    Description, as the server wrote it
    [FREE] Shrink a JSON document for LLM context: drop null/empty values, truncate long arrays and strings (with counts), limit depth, optional key allowlist. Reports estimated tokens saved. Limit 120/hour per IP.

    Annotations the server declares

    • title: Trim big JSON (free)
    • readOnlyHint: true
    • destructiveHint: false
    • idempotentHint: true
    • openWorldHint: false
  • url_to_markdown[FREE] Fetch a public web page and return only its main content as compact Markdown (headings, lists, links, images; scripts/styles/nav/header/cookie banners/footer/ads stripped).Reads text others writeCan send data outNo findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    • The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.

    What it can reach

    • untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
    • outboundThe input takes a URL (input.url): a request to a URL can carry data to whoever runs it.
    Description, as the server wrote it
    [FREE] Fetch a public web page and return only its main content as compact Markdown (headings, lists, links, images; scripts/styles/nav/header/cookie banners/footer/ads stripped). Does NOT execute JavaScript. JS-empty, challenge/blocked, invalid or private URLs return 4xx/5xx JSON with error.code, hint, charged:false. Reports original vs output size and estimated tokens saved. No LLM. Limit 30/hour per IP.

    Annotations the server declares

    • title: URL to compact Markdown (free)
    • readOnlyHint: true
    • destructiveHint: false
    • idempotentHint: true
    • openWorldHint: true
  • util_hash[PAID $0.005 USDC per call via x402 on Base; errors are free] UTF-8 text (GET ?text= or POST JSON {"text": "..."}, max 100000 chars) hashed with SHA-256, SHA-512, SHA3-256 and kecc…Reads text others writeNo findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    • The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.

    What it can reach

    • untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
    Description, as the server wrote it
    [PAID $0.005 USDC per call via x402 on Base; errors are free] UTF-8 text (GET ?text= or POST JSON {"text": "..."}, max 100000 chars) hashed with SHA-256, SHA-512, SHA3-256 and keccak256 (0x-prefixed, as used by Ethereum). Missing/oversized text = 400, never charged.

    Annotations the server declares

    • readOnlyHint: true
    • destructiveHint: false
    • idempotentHint: true
    • openWorldHint: true
  • util_random[PAID $0.005 USDC per call via x402 on Base; errors are free] CSPRNG (crypto.getRandomValues) with rejection sampling (no modulo bias).Reads text others writeNo findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    • The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.

    What it can reach

    • untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
    Description, as the server wrote it
    [PAID $0.005 USDC per call via x402 on Base; errors are free] CSPRNG (crypto.getRandomValues) with rejection sampling (no modulo bias). Optional min/max (default 0..1000000, range <= 2^32) and bytes 1-256 (default 32). Invalid input = 400, never charged.

    Annotations the server declares

    • readOnlyHint: true
    • destructiveHint: false
    • idempotentHint: true
    • openWorldHint: true
  • util_time[PAID $0.005 USDC per call via x402 on Base; errors are free] Authoritative current time from the Cloudflare edge clock.Reads text others writeNo findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    • The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.

    What it can reach

    • untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
    Description, as the server wrote it
    [PAID $0.005 USDC per call via x402 on Base; errors are free] Authoritative current time from the Cloudflare edge clock. Optional tz (IANA name, e.g. Asia/Saigon, default UTC). Unknown timezone = 400, never charged.

    Annotations the server declares

    • readOnlyHint: true
    • destructiveHint: false
    • idempotentHint: true
    • openWorldHint: true
  • util_uuid[PAID $0.005 USDC per call via x402 on Base; errors are free] RFC 9562 version 4 UUIDs from crypto.randomUUID at the edge.Reads text others writeNo findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    • The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.

    What it can reach

    • untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
    Description, as the server wrote it
    [PAID $0.005 USDC per call via x402 on Base; errors are free] RFC 9562 version 4 UUIDs from crypto.randomUUID at the edge. Optional count 1-100 (default 1).

    Annotations the server declares

    • readOnlyHint: true
    • destructiveHint: false
    • idempotentHint: true
    • openWorldHint: true
  • website_audit[PAID $0.05 USDC per call via x402 on Base; errors are free] Pre-launch website audit: up to 3 pages crawled, 30 links checked, meta/OG/canonical, security headers, HTTPS, 404 hand…Reads text others writeCan send data outNo findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    • The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.

    What it can reach

    • untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
    • outboundThe input takes a URL (input.url): a request to a URL can carry data to whoever runs it.
    Description, as the server wrote it
    [PAID $0.05 USDC per call via x402 on Base; errors are free] Pre-launch website audit: up to 3 pages crawled, 30 links checked, meta/OG/canonical, security headers, HTTPS, 404 handling, sitemap, and leaked-secret patterns in client JS (masked). Read-only.

    Annotations the server declares

    • title: Website pre-launch audit
    • readOnlyHint: true
    • destructiveHint: false
    • idempotentHint: true
    • openWorldHint: true
  • x402_preflight[PAID $0.005 USDC per call via x402 on Base; errors are free] For developers of buyer agents.Reads text others writeCan send data out1 finding

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    • The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.

    What it can reach

    • untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
    • outboundThe input takes a URL (input.url): a request to a URL can carry data to whoever runs it.

    Findings

    • Info

      The description mentions "credentials".

      Matched text

      description: "...y; IPs, localhost, .local/.internal and credentials refused (400). Unreachable/timeout targ..."

      What this check looks for
    Description, as the server wrote it
    [PAID $0.005 USDC per call via x402 on Base; errors are free] For developers of buyer agents. Makes ONE unpaid request (GET, POST fallback if GET is not allowed; 4 s timeout; max 2 redirects; 64 KB read) to the target and parses its HTTP 402 challenge (v2 PAYMENT-REQUIRED base64 header or v1 JSON body). Returns normalised accepts[] (scheme, CAIP-2 network, asset, amount in USD, payTo), pass/fail checks (valid challenge, supported mainnet network, USDC asset, parseable amount within a sanity cap, well-formed payTo, latency, TLS, Bazaar metadata, CORS), a verdict with plain reasons, latency, and the measured /ai-search score + sample size when the host is in our index. NEVER sends a payment or payment header. https + public hostnames + port 443 only; IPs, localhost, .local/.internal and credentials refused (400). Unreachable/timeout target = 502/504 JSON with verdict do_not_pay, never charged.

    Annotations the server declares

    • title: x402 pre-flight check (will this URL accept payment, what it costs)
    • readOnlyHint: true
    • destructiveHint: false
    • idempotentHint: true
    • openWorldHint: true
Only reads 5Show 5 read-only tools
  • base_snapshot[PAID $0.003 USDC per call via x402 on Base; errors are free] Bundle for agents that check Base before acting: everything in /api/base/gas (base fee, priority-fee tiers, trend, con…Reads text others writeNo findings

    Why it is in this group

    • The name "base_snapshot" says "snapshot": a word for reading.

    What it can reach

    • untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
    Description, as the server wrote it
    [PAID $0.003 USDC per call via x402 on Base; errors are free] Bundle for agents that check Base before acting: everything in /api/base/gas (base fee, priority-fee tiers, trend, congestion, USD cost of ETH transfer / ERC-20 transfer / swap incl. L1 data fee) + ETH/USD from Chainlink on Base, Coinbase spot and CoinGecko (median price, max-min deviationPct, 24h change) + network status: ok/degraded/stalled/sequencer-down from head block age, safe and finalized lag in blocks, and the Chainlink L2 sequencer uptime feed. One-line `summary` included. No params. Cached 5 s. Null = source unavailable (never invented). Errors are never charged.

    Annotations the server declares

    • title: Base snapshot: gas + ETH price + network status
    • readOnlyHint: true
    • destructiveHint: false
    • idempotentHint: false
    • openWorldHint: true
  • base_tx_status[PAID $0.002 USDC per call via x402 on Base; errors are free] Look up one transaction on Base mainnet by hash: status (success, failed or pending), block, timestamp, confirmations,…Reads text others writeNo findings

    Why it is in this group

    • The name "base_tx_status" says "status": a word for reading.

    What it can reach

    • untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
    Description, as the server wrote it
    [PAID $0.002 USDC per call via x402 on Base; errors are free] Look up one transaction on Base mainnet by hash: status (success, failed or pending), block, timestamp, confirmations, from/to, value, method id, gas used, effective gas price, fee split into L2 execution and L1 data fee (ETH and USD), contract created, and decoded ERC-20 Transfer events (token, symbol, from, to, amount; up to 20). Useful to confirm a payment landed. Read live from public Base RPC. All params optional: defaults to a confirmed USDC x402 payment. Unknown hash = 404, invalid hash = 400, never charged.

    Annotations the server declares

    • title: Base transaction status
    • readOnlyHint: true
    • destructiveHint: false
    • idempotentHint: true
    • openWorldHint: true
  • compress_log[FREE] Compress log output: strip timestamps/ids noise, dedupe repeated lines with (xN) counts, collapse stack traces, keep head/tail and all error/warn lines within max_lines.No findings

    Why it is in this group

    • The name "compress_log" says "log": a word for reading.
    Description, as the server wrote it
    [FREE] Compress log output: strip timestamps/ids noise, dedupe repeated lines with (xN) counts, collapse stack traces, keep head/tail and all error/warn lines within max_lines. Reports estimated tokens saved. Limit 120/hour per IP.

    Annotations the server declares

    • title: Compress logs (free)
    • readOnlyHint: true
    • destructiveHint: false
    • idempotentHint: true
    • openWorldHint: false
  • screenshot_url[PAID $0.008 USDC per call via x402 on Base; errors are free] Render a public web page in a real headless Chrome (JavaScript executed, waits for network idle) and return a screensh…Reads text others writeCan send data outNo findings

    Why it is in this group

    • The name "screenshot_url" says "screenshot": a word for reading.

    What it can reach

    • untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
    • untrusted-inputThe name "screenshot_url" says "screenshot" and "url": it reads web pages, which anyone can write.
    • untrusted-inputThe description's first sentence says "web": it reads web pages, which anyone can write.
    • untrusted-inputThe input takes a URL (input.url), and the tool reads it: it reads web pages, which anyone can write.
    • outboundThe input takes a URL (input.url): a request to a URL can carry data to whoever runs it.

    Claude Code: ask, because it is tagged outbound.

    Description, as the server wrote it
    [PAID $0.008 USDC per call via x402 on Base; errors are free] Render a public web page in a real headless Chrome (JavaScript executed, waits for network idle) and return a screenshot: viewport (default 1280x800) or full page (up to 8000 px tall), PNG or JPEG, optional dark mode and extra wait. JSON with image_base64 + finalUrl/status/title, or raw image bytes with format=png|jpeg. Typically 2-6 s (up to ~25 s when busy). Use it for visual QA, link previews, archiving, checking what a page really shows, or feeding a vision model. Adult/malware domains (Cloudflare family DNS) and private addresses are refused. All params optional: with none it captures https://example.com. Errors are never charged.

    Annotations the server declares

    • title: Screenshot a URL (headless Chrome)
    • readOnlyHint: true
    • destructiveHint: false
    • idempotentHint: true
    • openWorldHint: true
  • summarize[PAID $0.02 USDC per call via x402 on Base; errors are free] Fetch a public web page (or take raw text) and return a structured JSON brief made by Llama 3.3 70B: title, language, c…Reads text others writeCan send data outNo findings

    Why it is in this group

    • The name "summarize" says "summarize": a word for reading.

    What it can reach

    • untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
    • untrusted-inputThe description's first sentence says "web": it reads web pages, which anyone can write.
    • untrusted-inputThe input takes a URL (input.url), and the tool reads it: it reads web pages, which anyone can write.
    • outboundThe input takes a URL (input.url): a request to a URL can carry data to whoever runs it.

    Claude Code: ask, because it is tagged outbound.

    Description, as the server wrote it
    [PAID $0.02 USDC per call via x402 on Base; errors are free] Fetch a public web page (or take raw text) and return a structured JSON brief made by Llama 3.3 70B: title, language, contentType, summary, keyPoints, named entities with types, topic tags, concrete facts (prices, dates, versions, limits) as label/value pairs, sentiment and audience. Grounded in the page only, no invented facts. Reads up to 24,000 chars of main content (2 MB HTML, 10 s fetch timeout). All params optional: with none it summarizes https://example.com. GET with query params, or POST a JSON body for long text (up to 200,000 chars accepted, first 24,000 analyzed).

    Annotations the server declares

    • title: Summarize URL or text to JSON brief
    • readOnlyHint: true
    • destructiveHint: false
    • idempotentHint: true
    • openWorldHint: true

This is not a safety rating, and no finding is not a safety guarantee. The groups and findings come from fixed checks of the tool definitions the server returned on the date shown, not from its code or what it does when called. How the checks work.

Exfil surface

Which tools read private data, which read text someone else can write, and which can send data out. One of each is a path a prompt injection can use.

No private-data, untrusted-input, outbound path in this server alone: no tool is tagged private-data.

Reads private data0
Reads files, mail, messages, databases, secrets, calendars or repositories.
Reads text others write37
Brings in text someone else can write: web pages, inbound email, messages, issues and comments.
Can send data out12
Can carry data to people or systems: it sends, posts or publishes, or takes a recipient, channel or URL.

Other servers in the same agent add their own tools to these paths. Audit your whole setup to see them together.

Remote posture

What anyone on the internet sees at the remote endpoint: an anonymous initialize and tools/list, and the public OAuth metadata.

Tools listed without sign-in

Looked at Nov 20, 2026, 18:53 UTC, unchanged since Nov 6, 2026. No credentials were sent and no tool was called. What is looked at.

Reachable
yes
HTTP status of an anonymous initialize
200
tools/list served without credentials
yes
Tools that change things, listed without credentials
35
Protocol version
2025-06-18
WWW-Authenticate on 401
not observed
resource_metadata in WWW-Authenticate
not observed
RFC 9728 protected resource metadata
not observed
Authorization servers
not observed
Legacy HTTP+SSE transport
no
Dynamic client registration
not observed
Client ID metadata documents
not observed
  • HighThe whole server

    The endpoint served tools/list without credentials, and the list has 35 tools that change things (2 irreversible, 33 write). Whether a call also works without credentials was not tested: tools are never called.

    Matched text

    https://quickbuild.quynguyenbac.workers.dev/mcp: tools/list returned 40 tools to an anonymous client, including translate (irreversible), fx_rate (irreversible), launchcheck_full (write), website_audit (write), page_metadata (write) and 30 more

    What this check looks for

Posture changes

  1. Nov 6, 2026
    • Tools that change things, listed without credentials: 31 to 35
  2. Oct 29, 2026
    • Tools that change things, listed without credentials: 15 to 31

Findings

Matches of fixed checks, with the text they matched.

1 high6 warn1 info
  • The description contains "ignore previous instructions", an instruction to set earlier instructions aside.

    Matched text

    description: "...promise): flags instruction overrides ("ignore previous instructions"), role/system impersonation (<|im_star..."

    What this check looks for
  • The server marks "translate" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".

    Matched text

    annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":true,"openWorldHint":true}; The input addresses people (input.to): what it sends cannot be taken back.

    What this check looks for
  • The server marks "generate_image" read-only (readOnlyHint: true), and its own name, description or input give the class "write".

    Matched text

    annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":true}; The name "generate_image" says "generate": a word for a change.

    What this check looks for
  • The input schema at properties.address.description contains a 42-character base64 string.

    Matched text

    inputSchema.properties.address.description: "...tional: defaults to 0xd8dA6BF26964aF9D7eEd9e03E53415D37aA96045 (vitalik.eth)."

    What this check looks for
  • The input schema at properties.address.description contains a 42-character base64 string.

    Matched text

    inputSchema.properties.address.description: "...l: defaults to USDC 0x833589fCD6eDb6E08f4c7C32D4f71b54bdA02913."

    What this check looks for
  • The server marks "fx_rate" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".

    Matched text

    annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":true,"openWorldHint":true}; The input addresses people (input.to): what it sends cannot be taken back.

    What this check looks for
Show 2 more findings
  • The server marks "extract_links_and_headings" read-only (readOnlyHint: true), and its own name, description or input give the class "write".

    Matched text

    annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":true,"openWorldHint":true}; The name "extract_links_and_headings" says "link": a word for a change.

    What this check looks for
  • The description mentions "credentials".

    Matched text

    description: "...y; IPs, localhost, .local/.internal and credentials refused (400). Unreachable/timeout targ..."

    What this check looks for

History

Changes to its tool definitions and to its remote endpoint, newest first.

  1. Tools5 added, 6 changed.+5 added~6 changedView diff
  2. PostureRemote posture changed. Tools that change things, listed without credentials: 31 -> 35.View diff
    • Tools that change things, listed without credentials: 31 to 35
  3. PostureRemote posture changed. Tools that change things, listed without credentials: 15 -> 31.View diff
    • Tools that change things, listed without credentials: 15 to 31
  4. Tools16 added, 1 changed.+16 added~1 changedView diff
  5. Tools1 added.+1 addedView diff

Install

Pinned to version 1.2.0. Values in angle brackets are yours to fill in.

claude mcp add --transport http agent-tools https://quickbuild.quynguyenbac.workers.dev/mcp
claude mcp add --transport http agent-tools https://quickbuild.quynguyenbac.workers.dev/mcp

Run in a terminal.

Claude Code permissions

Allow the reads, ask before a change or a way out, deny the deletes. Paste into .claude/settings.json.

.claude/settings.json
{
  "permissions": {
    "allow": [
      "mcp__agent-tools__base_snapshot",
      "mcp__agent-tools__base_tx_status",
      "mcp__agent-tools__compress_log"
    ],
    "ask": [
      "mcp__agent-tools__launchcheck_full",
      "mcp__agent-tools__website_audit",
      "mcp__agent-tools__page_metadata",
      "mcp__agent-tools__domain_intel",
      "mcp__agent-tools__x402_preflight",
      "mcp__agent-tools__injection_check",
      "mcp__agent-tools__summarize",
      "mcp__agent-tools__translate",
      "mcp__agent-tools__generate_image",
      "mcp__agent-tools__screenshot_url",
      "mcp__agent-tools__text_to_speech",
      "mcp__agent-tools__base_gas_price",
      "mcp__agent-tools__base_wallet_balances",
      "mcp__agent-tools__base_token_info",
      "mcp__agent-tools__base_token_batch",
      "mcp__agent-tools__base_block",
      "mcp__agent-tools__base_name",
      "mcp__agent-tools__base_risk",
      "mcp__agent-tools__base_activity",
      "mcp__agent-tools__sol_wallet",
      "mcp__agent-tools__sol_token",
      "mcp__agent-tools__address_brief",
      "mcp__agent-tools__util_time",
      "mcp__agent-tools__util_uuid",
      "mcp__agent-tools__util_random",
      "mcp__agent-tools__util_hash",
      "mcp__agent-tools__base_token_verdict",
      "mcp__agent-tools__llm_chat",
      "mcp__agent-tools__text_embeddings",
      "mcp__agent-tools__rerank_documents",
      "mcp__agent-tools__crypto_price",
      "mcp__agent-tools__fx_rate",
      "mcp__agent-tools__free_site_check",
      "mcp__agent-tools__url_to_markdown",
      "mcp__agent-tools__trim_json",
      "mcp__agent-tools__estimate_tokens",
      "mcp__agent-tools__extract_links_and_headings"
    ],
    "deny": []
  }
}

Tools that only read are allowed, tools that change things ask first, tools that delete or cancel are denied. A read that can send data out (it takes a URL, say) asks first too. It assumes the server is installed as agent-tools. Review it before you use it.

Capability tags behind 37 rules
  • mcp__agent-tools__launchcheck_full untrusted-input, outbound
  • mcp__agent-tools__website_audit untrusted-input, outbound
  • mcp__agent-tools__page_metadata untrusted-input, outbound
  • mcp__agent-tools__domain_intel untrusted-input
  • mcp__agent-tools__x402_preflight untrusted-input, outbound
  • mcp__agent-tools__injection_check untrusted-input, outbound
  • mcp__agent-tools__summarize untrusted-input, outbound
  • mcp__agent-tools__translate untrusted-input, outbound
  • mcp__agent-tools__generate_image untrusted-input
  • mcp__agent-tools__screenshot_url untrusted-input, outbound
  • mcp__agent-tools__text_to_speech untrusted-input
  • mcp__agent-tools__base_gas_price untrusted-input
  • mcp__agent-tools__base_wallet_balances untrusted-input
  • mcp__agent-tools__base_token_info untrusted-input
  • mcp__agent-tools__base_token_batch untrusted-input
  • mcp__agent-tools__base_snapshot untrusted-input
  • mcp__agent-tools__base_tx_status untrusted-input
  • mcp__agent-tools__base_block untrusted-input
  • mcp__agent-tools__base_name untrusted-input
  • mcp__agent-tools__base_risk untrusted-input
  • mcp__agent-tools__base_activity untrusted-input
  • mcp__agent-tools__sol_wallet untrusted-input
  • mcp__agent-tools__sol_token untrusted-input
  • mcp__agent-tools__address_brief untrusted-input
  • mcp__agent-tools__util_time untrusted-input
  • mcp__agent-tools__util_uuid untrusted-input
  • mcp__agent-tools__util_random untrusted-input
  • mcp__agent-tools__util_hash untrusted-input
  • mcp__agent-tools__base_token_verdict untrusted-input
  • mcp__agent-tools__llm_chat untrusted-input
  • mcp__agent-tools__text_embeddings untrusted-input
  • mcp__agent-tools__rerank_documents untrusted-input
  • mcp__agent-tools__crypto_price untrusted-input
  • mcp__agent-tools__fx_rate untrusted-input, outbound
  • mcp__agent-tools__free_site_check untrusted-input, outbound
  • mcp__agent-tools__url_to_markdown untrusted-input, outbound
  • mcp__agent-tools__extract_links_and_headings untrusted-input, outbound

Pin

Keep the decision: fail CI when the tools change, follow every change, or show a badge.

Pin in CI

Fail a build when the tools you reviewed are not the tools the server serves today.

curl -s "https://radar.anlyon.com/api/v1/servers/dev.workers.quynguyenbac.quickbuild/agent-tools/verify?hash=0dde5cd7587f6b05f4eead470d4e9f156410c843cd44559f665287cb76edcff8" | jq -e '.changed == false'

Pinned hash 0dde5cd7587f, from Nov 10, 2026, 22:18 UTC.

Watch

Get told when its tools change: a new tool, a new description, a new class.

https://radar.anlyon.com/servers/dev.workers.quynguyenbac.quickbuild/agent-tools/changes.xml

An Atom feed. Works in any feed reader or Slack's RSS app.

Badge

Anlyon Radar badge for dev.workers.quynguyenbac.quickbuild/agent-tools
[![Anlyon Radar](https://radar.anlyon.com/badge/dev.workers.quynguyenbac.quickbuild/agent-tools.svg)](https://radar.anlyon.com/servers/dev.workers.quynguyenbac.quickbuild/agent-tools)

Source: MCP Registry. Synced Oct 29, 2026, 13:17 UTC.

QuickBuild token-saver tools + x402 MCP server: questions and answers

Answered from the scan above, not from the publisher's description.

Is the QuickBuild token-saver tools + x402 MCP server safe?

No scan can promise that. What its definitions show: 40 tools, 35 of which can change, send or delete something, and 8 findings (1 high) from fixed checks. Allow the reads, ask before each change, and deny the deletes; the permissions block on this page does that.

Can the QuickBuild token-saver tools + x402 MCP server delete data?

No tool is classed as destructive: going by their definitions, none of them deletes, cancels or revokes anything.

Can the QuickBuild token-saver tools + x402 MCP server do things that cannot be undone?

Yes. 2 tools can send, pay, publish or merge: translate and fx_rate.

What tools does the QuickBuild token-saver tools + x402 MCP server have?

40 tools: 2 irreversible, 33 write, 5 read. translate, fx_rate, launchcheck_full, website_audit, page_metadata, domain_intel, x402_preflight, injection_check and 32 more.

Does the QuickBuild token-saver tools + x402 MCP server need sign-in or OAuth?

Its remote endpoint listed its tools to an anonymous client, without sign-in.

Does the QuickBuild token-saver tools + x402 MCP server send data to outside services?

12 tools can carry data out (it sends, posts or publishes, or takes a recipient, channel or URL): launchcheck_full, website_audit, page_metadata, x402_preflight, injection_check, summarize and 6 more.

Has the QuickBuild token-saver tools + x402 MCP server changed its tools?

Yes, 3 times since Anlyon Radar started tracking it, most recently on Nov 10, 2026. Every change is on this page and in its Atom feed.