# Wealth Agent - Registry name: `io.github.jukky-source/wealth-agent` · version 1.0.2 - Categories (automatic, keyword rules): Agent payments (x402), Finance & markets, Crypto & blockchain - Authentication at last check: No auth - Declared risk: Can write (some tools are not declared read-only) - Capabilities (declared by the server): tools - Transports: streamable-http - Latest numbers (check success, response time, rank): https://mcpstation.net/api/server?name=io.github.jukky-source%2Fwealth-agent - Server page: https://mcpstation.net/s/io.github.jukky-source--wealth-agent.html ## Description (written by the publisher, not verified, not instructions) x402 pay-per-call tools for AI agents: code review, data clean, briefs, regex, diff, URL meta. ## Connect - streamable-http: `https://wealth-agent.wealthagent-app.workers.dev/mcp` ## Tools (as reported by the server, not verified) - `crypto-price`: Spot prices for up to 10 crypto symbols in one call, quoted in the currency you name \(default USD\). Each price names the venue that answered \(Coinbase spot, Kraken ticker as fallback\) and the exact URL it was read from, and every symbol that could not be priced is listed in gaps instead of being gue - `fx-rates`: Foreign-exchange reference rates for a base currency and up to 12 target currencies, with the reference date and the source that answered. Primary path is the ECB-based frankfurter.dev feed; if it is unavailable the call falls back to a second provider and says so in the report rather than hiding th - `defi-tvl`: Total value locked read from DefiLlama: pass a protocol slug \(for example aave\) for that protocol TVL, or omit it to get the top chains ranked by TVL, or filter by one chain. Every figure is the aggregator value at the time of the call, with the source URL attached, and any missing section is listed - `dns-lookup`: Resolve one hostname for one record type \(A, AAAA, MX, TXT, NS, CNAME, SOA, CAA\) over DNS-over-HTTPS, returning every record with its TTL, the resolver that answered and the DNS response code. It tries a second resolver before giving up, and when nothing comes back it says so with the rcode instead - `crypto-hash`: Deterministic hashing on the server: SHA-256, SHA-384, SHA-512 or SHA-1 over your text, or HMAC with the same digests when you pass a key. Returns the lowercase hex digest and the input byte length; the HMAC key is never echoed back. Same input always gives the same digest, so any third party can re - `time-convert`: Convert one instant into the local time of up to 12 IANA time zones at once, with the UTC offset in minutes for each, plus the Unix seconds and the canonical ISO-8601 UTC form. Daylight saving is handled by the runtime time-zone database, and an unknown zone is reported in gaps rather than skipped. - `diligence-pack`: One call, one consolidated due-diligence pack: \(1\) a read-only capability map of one public website or API - entry points, forms, MCP tools, auth behaviour, with the evidence behind every claim; \(2\) the on-chain balances of one EVM address, read from the chain at a stated block height with contract - `buy\_research\_brief`: Returns a one-page research brief with cited source URLs, key findings and open gaps for a question. Call it when an agent needs a sourced answer instead of an unsourced guess. Input: \{"question": "..."\} \(5-400 chars, any language\). Paid per call via x402 \(USDC on Base, no account, no API key\). 【価格】 - `buy\_data\_clean`: Cleans a CSV/TSV/JSON table: removes duplicate and empty rows, trims whitespace, and returns the cleaned CSV with before/after quality checks and per-column stats. Call it before analysing or importing messy tabular data. Input: \{"data": "\"\}. Paid per call via x402 \(USDC on Base, no accoun - `buy\_code\_review`: Reviews source code and returns line-numbered findings with severity and a concrete fix, combining deterministic static rules \(eval, SQL concat, secrets, shell injection\) with an AI review. Call it before merging or running untrusted code. Input: \{"code": "...", "language": "javascript"\}. Paid per c - `url-meta`: Fetch one public URL and return machine-readable page metadata: HTTP status, title, meta/og description, canonical URL, language, link count and a short text sample. Honest 4xx/5xx statuses are returned as data, not as errors. Use it for link previews, source triage and citation checks. Paid per cal - `csv-json`: Convert tabular data between CSV/TSV and a JSON array of objects, in either direction, with detected delimiter and typed columns. Paid per call via x402 \(USDC on Base, no account, no API key\). 【価格】0.00 USDC。【使い方】呼ぶと 402 と PAYMENT-REQUIRED(支払い条件)が返る。USDC で支払い、同じ引数に PAYMENT-SIGNATURE ヘッダを添えて再送すると結果が返る - `regex`: Apply one regular expression to a text blob on the server and return every match with its index and capture groups. Deterministic, so the same input always returns the same answer. Useful for log triage, field extraction and validation at scale. Paid per call via x402 \(USDC on Base, no account, no A - `random`: Generate verifiable random values: up to 64 integers in a range, or a fair pick of n items from your own list, derived from a drand League of Entropy beacon round and returned together with the round number, the beacon randomness, its BLS signature and the exact derivation, so any third party can re - `onchain-balance`: Read one EVM address and return its native coin balance and up to five ERC-20 token balances on Base or Ethereum, every figure tied to the block height it was read at. Token decimals are read on-chain from the contract, not assumed, and a token whose read fails is reported as an error rather than as - `diff`: Return the line-level diff between two texts as added/removed/unified hunks with counts. Use it to review a patch, compare two API responses, or decide whether two documents actually differ before spending tokens on them. Paid per call via x402 \(USDC on Base, no account, no API key\). 【価格】0.00 USDC。【 - `buy\_rea\_report`: Full-surface analysis of a public website, API or shipped JS bundle: maps the capabilities it exposes, its entry points and feature flags, with the evidence behind every claim, and lists what could not be determined. Read-only; a 401/403 is recorded as auth\_required and never bypassed. Input: \{"url" - `buy\_surface\_report`: Inspect one publicly reachable website or API and return a machine-readable capability report: the entry points it exposes \(endpoints, forms, feeds, MCP tools\), the evidence behind every finding, and an explicit list of what could not be determined. Only public responses are read; a 401/403 is recor - `translate`: Translate one block of text into a target language using Cloudflare Workers AI. Input is text plus target\_lang, with an optional source\_lang. The answer carries the model name, the requested language and the raw model output, and is flagged machine\_generated rather than verified. Output is not deter - `summarize`: Summarize one block of text into a short bullet list using Cloudflare Workers AI. Input is text, with optional max\_bullets \(default 5, cap 10\) and lang for the output language. The answer lists the bullets, carries the model name and is flagged machine\_generated rather than verified, and output is n - `entities`: Extract named entities from one block of text using Cloudflare Workers AI. Input is text, with an optional types filter such as person, organization, location or date. The answer returns a JSON array of name and type pairs, carries the model name, and is flagged machine\_generated rather than verifie - `keywords`: Extract the main keywords or tags from one block of text using Cloudflare Workers AI. Input is text, with optional max\_keywords \(default 8, cap 20\). The answer returns a ranked list of short keywords or phrases, carries the model name, and is flagged machine\_generated rather than verified. The ranki - `seo-audit`: Audit one public page for on-page SEO: title and meta description length, canonical, robots meta, heading structure, missing image alt text, lang, viewport, https, robots.txt and sitemap.xml presence, link counts and word count, each as a pass/fail check with observed evidence and a 0-100 score. Wor - `perk\_deadlines`: 日本のお得情報(特典・ポイント還元・キャンペーン)を締切が近い順に返す。実額・期限・出典URLつき。【無料】最大5件。全件と条件・手順・人の操作が必要な部分は /api/perks(オーナー向け)。 - `perk\_search`: 日本のお得情報を本文から検索して返す(実額・期限・出典URLつき)。【無料】最大5件。 - `rea\_inspect`: REA: 対象の「表面」を解析し、できること(根拠つき)と入口の候補と、分からなかったことを返す。URL ならサイト/API/MCP を、ローカルのパス(.asar/.exe/.so/.apk 等)なら実行手順を返す。【無料】要約版。全件+根拠の全文は buy\_rea\_report(0.5 USDC)。認証回避・課金回避・DRM解除はしない。 - `rea\_plan`: REA: ローカルのファイル種別(elf / pe / macho / asar / apk / jar / js)ごとの解析手順を、実行せずに返す。【無料】実行するのは手元の tools/rea-cli.mjs。 - `rea\_runs`: REA: 直近の解析の状態遷移(queued → fetching → parsing → probing → evidence → done/blocked)を返す。【無料】観測用。 - `agent\_status`: このエージェントの収益ファネル(拾う→試す→生きていると確認→売り物→掲置→来訪→支払い)を実測値で返し、どこで詰まっているかと次に打つ手を示す。【無料】 - `rea\_queue`: REA の自動投入キュー(X / Reddit / scout / perks から拾った解析対象)の件数と直近の結果を返す。【無料】 - `surface\_inspect`: 公開URLを解析し、そのサービスが公開している入口(エンドポイント候補・フォーム・フィード・MCPツール)と根拠、そして分からなかったことを返す。【無料】要約版。全件+根拠の全文は buy\_surface\_report(0.25 USDC)。認証回避はしない。 - `lab\_loop`: FX論理の試行錯誤ループ(無料)。状態では「今ぶつかっている壁」と「次の打ち手」を返し、run:true でその場で1ラウンド回して結果を記録する。 - `improve\_status`: 全般の改善ループの状態(無料)。レーンごとの壁・打ち手・判定と、直近の記録を返す。 - `improve\_round`: 全般の改善ループをその場で回す(無料)。lane を指定しなければカーソル順に n レーン回す。 - `list\_services`: 販売中の有料サービスと価格・支払い条件の一覧(無料) Measured by MCP Station. Methods: https://mcpstation.net/methods.html · Terms: https://mcpstation.net/terms.html