Doc Extract API
What it does
SchemaLock — paid extraction: JSON from docs via your schema, x402/USDC.
- Connects 在你的电脑上运行
- Account or key Needs WALLET_PRIVATE_KEY
- Category 开发者工具
安装
选择你使用的 AI 应用,然后复制设置。
通过 3/3 项检查 · 在你的电脑上运行 · ✓ 命令已检查 2026年11月13日
第一步:你用哪个 AI?
选择你用的应用,步骤会显示在下方。
注册并创建 Doc Extract API 的密钥
打开开发者网站,注册账户并创建 API 密钥,下一步要用。价格和免费方案请看开发者网站。
获取密钥的方法见下方开发者的 README
打开电脑版 Claude 应用的设置文件
SettingsDeveloperEdit Config
粘贴这些设置并保存文件
{ "mcpServers": { "doc-extract-api": { "command": "npx", "args": [ "-y", "doc-extract-api" ], "env": { "WALLET_PRIVATE_KEY": "your-value" } } } }只把密钥粘贴到应用的输入框里,不要粘贴到聊天中或发给任何人
退出 Claude 再重新打开
试一试
用 Doc Extract API 告诉我你能帮我做什么,然后做一件给我看✓ 如果 Claude 说正在使用 Doc Extract API,或回答中出现 Doc Extract API 的工具名称,就表示连接成功
菜单名称已于 2026年11月11日 对照 Anthropic 帮助页面核实 · 你的界面可能因版本略有不同
ChatGPT 暂时无法连接 Doc Extract API
Doc Extract API 需要安装在你的电脑上,而 ChatGPT 只能连接在网上运行的服务器
你可以改用这些方式
A · 使用能连接 Doc Extract API 的应用
- 需要 Doc Extract API 的密钥(在开发者网站创建) · 需要先在电脑上安装 Node.js(只需一次) · 仅限电脑上的 Claude 应用(Windows 或 Mac),网页和手机不可用
- 需要 Doc Extract API 的密钥(在开发者网站创建) · 需要先在电脑上安装 Node.js(只需一次)
- 需要 Doc Extract API 的密钥(在开发者网站创建) · 需要先在电脑上安装 Node.js(只需一次)
- 需要 Doc Extract API 的密钥(在开发者网站创建) · 需要先在电脑上安装 Node.js(只需一次)
B · 开发者工具 类中能连接 ChatGPT 的其他服务器
从同类中使用较多的里挑选,仅限用地址即可连接、无需密钥的
- Sentry✓ ChatGPT
- Compiler Explorer✓ ChatGPT
- GitDiagram✓ ChatGPT
可连接 Claude Code,但需要先准备
需要 Doc Extract API 的密钥(在开发者网站创建) · 需要先在电脑上安装 Node.js(只需一次)
注册并创建 Doc Extract API 的密钥
打开开发者网站,注册账户并创建 API 密钥,下一步要用。价格和免费方案请看开发者网站。
获取密钥的方法见下方开发者的 README
打开终端
Mac:按 Command+空格,输入 Terminal · Windows:打开 PowerShell
粘贴这条命令并按 Enter
claude mcp add --scope user doc-extract-api --env WALLET_PRIVATE_KEY=your-value -- npx -y doc-extract-api按 Enter 前,把“your-value”改成你的密钥
只把密钥粘贴到应用的输入框里,不要粘贴到聊天中或发给任何人
在 Claude Code 中输入 /mcp
应能看到 Doc Extract API 在列表中并显示已连接
试一试
用 Doc Extract API 告诉我你能帮我做什么,然后做一件给我看✓ 如果 Claude Code 说正在使用 Doc Extract API,或回答中出现 Doc Extract API 的工具名称,就表示连接成功
菜单名称已于 2026年11月11日 对照 Anthropic 帮助页面核实 · 你的界面可能因版本略有不同
可连接 Cursor,但需要先准备
需要 Doc Extract API 的密钥(在开发者网站创建) · 需要先在电脑上安装 Node.js(只需一次)
注册并创建 Doc Extract API 的密钥
打开开发者网站,注册账户并创建 API 密钥,下一步要用。价格和免费方案请看开发者网站。
获取密钥的方法见下方开发者的 README
填写你的密钥
在 Doc Extract API 的设置中,WALLET_PRIVATE_KEY 还是空的,填入你的密钥并保存
只把密钥粘贴到应用的输入框里,不要粘贴到聊天中或发给任何人
在 Cursor 中以 Agent 模式打开 AI 聊天
试一试
用 Doc Extract API 告诉我你能帮我做什么,然后做一件给我看✓ 如果 Cursor 说正在使用 Doc Extract API,或回答中出现 Doc Extract API 的工具名称,就表示连接成功
菜单名称已于 2026年11月11日 对照 Cursor 帮助页面核实 · 你的界面可能因版本略有不同
可连接 VS Code,但需要先准备
需要 Doc Extract API 的密钥(在开发者网站创建) · 需要先在电脑上安装 Node.js(只需一次)
注册并创建 Doc Extract API 的密钥
打开开发者网站,注册账户并创建 API 密钥,下一步要用。价格和免费方案请看开发者网站。
获取密钥的方法见下方开发者的 README
填写你的密钥
在 Doc Extract API 的设置中,WALLET_PRIVATE_KEY 还是空的,填入你的密钥并保存
只把密钥粘贴到应用的输入框里,不要粘贴到聊天中或发给任何人
在 VS Code 中以 Agent 模式打开 AI 聊天
试一试
用 Doc Extract API 告诉我你能帮我做什么,然后做一件给我看✓ 如果 VS Code 说正在使用 Doc Extract API,或回答中出现 Doc Extract API 的工具名称,就表示连接成功
菜单名称已于 2026年11月11日 对照 VS Code 帮助页面核实 · 你的界面可能因版本略有不同
开发者专用:终端命令和 JSON 设置
安装在电脑上
Claude Code
claude mcp add --scope user doc-extract-api --env WALLET_PRIVATE_KEY=your-value -- npx -y doc-extract-apiClaude Desktop (claude_desktop_config.json)
{
"mcpServers": {
"doc-extract-api": {
"command": "npx",
"args": [
"-y",
"doc-extract-api"
],
"env": {
"WALLET_PRIVATE_KEY": "your-value"
}
}
}
}Cursor (~/.cursor/mcp.json)
{
"mcpServers": {
"doc-extract-api": {
"command": "npx",
"args": [
"-y",
"doc-extract-api"
],
"env": {
"WALLET_PRIVATE_KEY": "your-value"
}
}
}
}VS Code (mcp.json)
{
"servers": {
"doc-extract-api": {
"type": "stdio",
"command": "npx",
"args": [
"-y",
"doc-extract-api"
],
"env": {
"WALLET_PRIVATE_KEY": "your-value"
}
}
}
}需要你填写的值
WALLET_PRIVATE_KEY机密 必填 envEVM private key for a wallet funded with real USDC on Base. Every tool call pays the live x402 API from this wallet — there is no free extraction path.
说明仍为英文原文。
这些设置是根据作者在注册中心的条目生成的。如果 README 的写法不同,请以 README 为准。
一览
哪些应用可以使用它
依据这个服务器的连接方式和各应用自己的文档,检查于 2026年11月10日。
在你的电脑上运行,适用于
ClaudeGooseLM StudioMsty StudioCherry StudioCursorVS Code + GitHub CopilotVisual StudioWindsurf (Devin Desktop)ZedKiroClineContinueJetBrains AI AssistantClaude CodeCodex CLIGemini CLIOpenCodenodecodeWarpLibreChatRaycast
它能做什么
它在你的电脑上运行;安装后,工具列表会显示在你的 AI 应用中。
基础检查i
我们检查的是任何人都能看到的信息;我们没有阅读代码。不适用的检查项不计入总数。 · 检查于 2026年11月14日
- ✓ 通过
作者与仓库一致
注册名称已通过 GitHub 账户 thestarboy9696 验证,该账户拥有此仓库
- ✓ 通过
仍在维护
代码最后更新于2个月前
- ✓ 通过
有许可证
使用 MIT
- – 不适用
端点有响应
它在你的电脑上运行,没有互联网端点
- – 不适用
使用 OAuth 登录
在你的电脑上运行,改用本地设置
- … 尚未检查
工具描述中没有隐藏指令
它在你的电脑上运行;我们只能读取互联网端点的工具列表
讨论
提问、分享技巧或报告无法使用。按“有帮助”数量排序。
用户评价
还没有评价
与“开发者工具”中的其他服务器比较i
- 星标数多于此分类中其他 5866 个服务器的 0%
- 发布到注册中心的时间比此分类中其他 5866 个服务器的 31% 更新
- 基础检查结果等于或优于此分类中其他 5866 个服务器的 100%
与注册中心里这个分类中所有活跃的服务器比较。只有当发布者拥有该仓库时,星标数才计入。
常见问题
Doc Extract API 安全吗?
它在 3 项适用的基础检查中通过了 3 项。我们没有阅读代码。在添加密钥或令牌之前,请先看看这个服务器要求哪些权限,并使用只具备所需权限的密钥。
它需要哪些密钥或代码?
你需要填写:WALLET_PRIVATE_KEY
它在我的电脑上运行还是在云端?
它在你的电脑上运行。你的 AI 应用会在需要时启动它,数据不会经过作者的服务器(除非它自己调用在线服务)。
它还在维护吗?
GitHub 仓库最后更新于2个月前,最新版本(0.2.0)于3个月前进入注册中心。
哪些应用可以使用它?
支持 MCP 的应用,例如 Claude Desktop、Claude Code、Cursor 和 VS Code。你的电脑上还需要有 Node.js。
同类别的其他选择
同一分类中其他作者的服务器,按星标数排序。
如何看卡片:↓ 每周下载量i★ GitHub 星标i三大支柱:评价 · 检查 · 使用情况i更新 = 最新版本发布到注册中心的时间i
GitHub
Connect AI assistants to GitHub - manage repos, issues, PRs, and workflows through natural language.
Serena MCP: the IDE for your agent
A powerful toolkit for coding, providing semantic retrieval and editing capabilities.
作者的 README
打开 README
按原文从作者的仓库复制。PopNx MCP 没有审查或背书其内容。HTML、图片和脚本已被移除。链接会离开本站。
doc-extract-api
Document-to-structured-JSON extraction API. Takes a PDF/PNG/JPEG and returns clean JSON. Four fixed document types (invoice, receipt, contract, resume) get a matched schema plus domain-specific validation; /extract/custom accepts *any* JSON Schema, so a caller isn't limited to those four types. Built for autonomous AI agents: no signup, no API key, no human checkout — an agent discovers the price and pays per call via x402 v2 (USDC), entirely over HTTP, and can find this service programmatically via the x402 Bazaar discovery catalog. Also published as an MCP server (doc-extract-api-mcp on npm, and on the official MCP Registry) — but that's a *client* for the same paid API, not a free alternative: every MCP tool call is a real x402 payment from a wallet you supply, not a locally-run, unmetered extraction. There is no free path to this service.
Why x402, not Stripe
Stripe Checkout is built for humans — it collects a card through a UI, and its payment button actively resists non-trusted, programmatic (agent-driven) submission. x402 revives the HTTP 402 Payment Required status code: hit a paid endpoint with no payment, get back a 402 whose PAYMENT-REQUIRED header carries a complete, machine-readable manifest (price, network, receiving address, asset, and — via the Bazaar extension — a full input/output schema with a real example). Retry with an X-PAYMENT header carrying a signed USDC transfer authorization, and the request goes through. No account, no key, no human in the loop — the discovery step and the payment step are the same HTTP round trip.
Payment is only settled after the extraction succeeds — @x402/hono's middleware verifies the payment signature before running the handler, but only executes the actual on-chain transfer if the handler returns a non-error status (confirmed: this project's paymentMiddlewareFromConfig call skips settlement whenever the wrapped route responds ≥400). A failed extraction costs the caller nothing.
Discovery: the x402 Bazaar
The Bazaar (GET {facilitator}/discovery/resources) is a live, queryable catalog of x402 services — ~15k listings as of this writing, searchable by agent clients (client.extensions.bazaar.search({ query: "..." })). Getting listed isn't a submission form: a facilitator catalogs a resource automatically the first time it processes a real payment that echoes the route's declared bazaar extension. src/lib/discovery.ts declares that extension for all four routes — full input schema (multipart form-data, a file field) and a real, verified output example — so once a route's first live mainnet payment lands, it becomes findable in the catalog with enough detail for an agent to integrate correctly without reading any docs. As of this writing the /extract/invoice catalog entry is still showing a stale pre-migration record rather than this rich content — see git history / ask for status; /extract/receipt, /extract/contract, and /extract/resume haven't had a mainnet payment yet so aren't catalogued at all.
Positioning (from surveying the live Bazaar before writing the listing copy): generic OCR competitors are free but return raw unvalidated text; generic structured-extraction competitors charge ~$0.10/call and require the caller to already have plain text, not a raw file. This is the only listing found doing both — raw PDF/image in, a fixed validated schema out. Pricing (see wrangler.toml) is matched per-route to the lowest verified live-402 price from a direct competitor in that category, not set independently — $0.02 for invoice/receipt/resume/custom, $0.03 for contract (its real comparable is a cent higher). See src/lib/discovery.ts for the exact listing copy and each route's tags (serviceName/tags/description).
Architecture
- `schemas/` — JSON Schema for
invoice,receipt,contract, andresumeextractions (also the toolinput_schemasent to Claude, so the model is forced into this exact shape). - `src/lib/claude.ts` — calls the Anthropic Messages API with the document as a native
document/imagecontent block and a forcedtool_choice, so the model can't return anything but the schema. - `src/lib/validate.ts` — post-extraction sanity checks: line items sum to subtotal, subtotal + tax (+ tip) ≈ total, dates are valid and not in the future, currency is a real ISO 4217 code, low-confidence fields get flagged.
- `src/lib/extract.ts` — orchestrates one extraction pass, validates, and — if validation fails — retries once with a corrective prompt describing exactly what didn't add up. Never fails silently; returns
validation_warningsalongside the data either way. - `src/lib/customSchema.ts` / `src/routes/customExtractHandler.ts` —
/extract/custom: same Claude forced-tool-use engine, but the tool'sinput_schemais a JSON Schema the caller supplies at request time instead of one of the four fixed schemas. There's no domain-specific validator (no fixed domain to check against), so this returns schema-conformant output, not fact-checked output —validation_warningsis always[]. The caller's schema is bounded (≤6000 chars, depth ≤4, ≤30 properties per level, no$ref) and rejected with a 400 — which costs nothing, since x402 only settles on a non-error response — before any model call. Accepts either afile(binary PDF/image, same as the fixed routes) orcontent(raw text/HTML the caller already has — e.g. an already-scraped web page, ≤100,000 chars) — exactly one of the two, not both. The text/HTML path is labeled explicitly as data in the prompt sent to Claude, not instructions, since it's arbitrary caller-supplied content and could otherwise be a prompt-injection vector. - `src/lib/facilitator.ts` — picks the free default facilitator on testnet, or, on mainnet (
X402_NETWORK = "base"), Coinbase's authenticated CDP facilitator plus Mogami and PayAI for additional Bazaar-equivalent discovery surface (CDP stays first/primary — earlier facilitators get precedence for verify/settle); maps this project's human-readable network names to the CAIP-2 identifiers x402 v2 uses (eip155:8453etc). Important:src/index.tscaches the built payment middleware at module scope rather than rebuilding it per-request — with multiple facilitators, an unhealthy one adds its full timeout toinitialize(), and rebuilding per-request meant paying that cost on *every* request (this broke the API entirely the first time multi-facilitator was tried). Caching means only a cold isolate's first request pays that cost. - `src/lib/discovery.ts` — the Bazaar listing content: descriptions, tags, and discovery extensions (schema + real example) for all four routes.
- `src/index.ts` — Hono app on Cloudflare Workers.
@x402/hono'spaymentMiddlewareFromConfigguards/extract/*, built per-request fromEnvbindings (Workers only exposes env vars inside a request, not at module load time). - `src/routes/extractHandler.ts` — pure extraction logic; payment has already been verified by the time this runs.
- `mcp/server.ts` — stdio MCP server exposing
extract_invoice,extract_receipt,extract_contract,extract_resume, andextract_customtools. Each tool call signs and sends a real x402 payment (via@x402/fetch'swrapFetchWithPayment, same mechanism asscripts/test-payment.mjs) from a wallet supplied viaWALLET_PRIVATE_KEY, then calls the live HTTP API over the network — it does not run extraction locally or call Claude directly. No free path: a missing or unfunded wallet fails the same way it would calling the API directly.
One deliberate deviation from the original spec
The spec assumed a PDF→image conversion step (pdf2image/poppler). Cloudflare Workers can't run native binaries, and Claude's Messages API accepts PDFs natively as a document content block — so PDFs are sent to Claude as-is. This removes a whole pipeline stage and a dependency that wouldn't run on Workers anyway.
Setup
cd "doc-extract-api"
npm install
1. Secrets
Local dev — copy .dev.vars.example to .dev.vars and fill in a real key (gitignored):
cp .dev.vars.example .dev.vars
Production:
npx wrangler secret put ANTHROPIC_API_KEY
2. Wallet — where payments land
Edit wrangler.toml: X402_PAY_TO_ADDRESS must be a real wallet address you control before this can receive actual payments. It's a public receiving address (safe to commit, like a bank account number) — never put a private key anywhere in this project.
X402_NETWORK defaults to base-sepolia (testnet, free fake USDC from a faucet — good for verifying the whole flow with zero real money at risk).
Pricing is per-route, not shared: X402_INVOICE_PRICE, X402_RECEIPT_PRICE, X402_CONTRACT_PRICE, X402_RESUME_PRICE, X402_CUSTOM_PRICE. Defaults are matched to the lowest verified live competitor per category — see the Positioning note above.
2b. Going to mainnet (X402_NETWORK = "base")
The free default facilitator (x402.org/facilitator, used automatically on testnet) only supports testnet — mainnet requires an authenticated facilitator. This project is wired for Coinbase's CDP facilitator (src/lib/facilitator.ts):
- Create a CDP Portal account and a Secret API Key under Project → API Keys.
npx wrangler secret put CDP_API_KEY_IDnpx wrangler secret put CDP_API_KEY_SECRET- Set
X402_NETWORK = "base"inwrangler.tomland redeploy.
Pricing: free for the first ~1,000 settled transactions/month, then $0.001 each — payment verification itself is always free. If X402_NETWORK is "base" and these secrets aren't set, requests will fail loudly at the facilitator-config step rather than silently misconfiguring payments.
Getting listed in the Bazaar requires at least one real mainnet payment to go through — the facilitator catalogs the resource the first time it processes a payment that echoes the bazaar extension. Run scripts/test-payment.mjs with a real funded wallet once to trigger it (see below), then confirm with:
curl -s "https://api.cdp.coinbase.com/platform/v2/x402/discovery/resources?limit=200" \
| grep -o 'doc-extract-api'
3. End-to-end payment test (scripts/test-payment.mjs)
Simulates a real autonomous agent: generates/uses a throwaway wallet, pays via x402 v2, and calls the live API. Fund a test wallet first — testnet USDC from a faucet (e.g. faucet.circle.com, Base Sepolia) for a free run, or a small amount of real USDC on Base if X402_NETWORK is "base":
node scripts/gen-test-wallet.mjs # prints a fresh throwaway private key + address
TEST_WALLET_PRIVATE_KEY=0x... node scripts/test-payment.mjs
Prints the settlement details and the recipient's before/after USDC balance — confirmed working end to end on both Base Sepolia (testnet) and Base mainnet with real USDC.
4. Run locally
npm run dev
curl -X POST http://localhost:8787/extract/invoice -F "file=@/path/to/invoice.pdf"
→ 402 Payment Required; full payment + discovery manifest in the PAYMENT-REQUIRED header
/extract/custom takes a schema field (a JSON Schema string, type: "object"), an optional instructions field, and exactly one of file (binary) or content (raw text/HTML):
curl -X POST http://localhost:8787/extract/custom \
-F "file=@/path/to/doc.pdf" \ -F 'schema={"type":"object","properties":{"order_id":{"type":"string"},"ship_date":{"type":"string"}}}' \ -F "instructions=Dates in YYYY-MM-DD"
Or from raw text/HTML you already have — no file needed:
curl -X POST http://localhost:8787/extract/custom \
-F "content= Acme Corp Founded 1998, HQ in Austi