AI Skill Hub 推荐使用:专利客户代理 是一款优质的Agent工作流。AI 综合评分 7.5 分,在同类工具中表现稳健。如果你正在寻找可靠的Agent工作流解决方案,这是一个值得深入了解的选择。
专利客户代理 是一套完整的 AI Agent 自动化工作流方案。通过可视化的节点编排,将复杂的多步骤任务拆解为清晰的自动化流程,实现全程无人值守的智能处理。支持与数百种外部服务和 API 无缝集成,适合构建数据处理管线、业务自动化和 AI 辅助决策系统。
专利客户代理 是一套完整的 AI Agent 自动化工作流方案。通过可视化的节点编排,将复杂的多步骤任务拆解为清晰的自动化流程,实现全程无人值守的智能处理。支持与数百种外部服务和 API 无缝集成,适合构建数据处理管线、业务自动化和 AI 辅助决策系统。
# 方式一:pip 安装(推荐)
pip install patent-client-agents
# 方式二:虚拟环境安装(推荐生产环境)
python -m venv .venv
source .venv/bin/activate # Windows: .venv\Scripts\activate
pip install patent-client-agents
# 方式三:从源码安装(获取最新功能)
git clone https://github.com/parkerhancock/patent-client-agents
cd patent-client-agents
pip install -e .
# 验证安装
python -c "import patent_client_agents; print('安装成功')"
# 命令行使用
patent-client-agents --help
# 基本用法
patent-client-agents input_file -o output_file
# Python 代码中调用
import patent_client_agents
# 示例
result = patent_client_agents.process("input")
print(result)
# patent-client-agents 配置文件示例(config.yml) app: name: "patent-client-agents" debug: false log_level: "INFO" # 运行时指定配置文件 patent-client-agents --config config.yml # 或通过环境变量配置 export PATENT_CLIENT_AGENTS_API_KEY="your-key" export PATENT_CLIENT_AGENTS_OUTPUT_DIR="./output"
<p align="center"> <a href="https://patentclient.com/atlas"> <img src="docs/_static/atlas_hero.png" alt="Patent Client Agents — worldwide coverage atlas" width="800"> </a> </p>
Give your AI agent access to the world's patent and trademark data.
Project home: patentclient.com · Full documentation: docs.patentclient.com
patent-client-agents is an MCP server, so it works with any MCP-speaking client — Claude Code, OpenAI Codex CLI, Google Gemini CLI, Cursor, Windsurf, Cline, Zed, Continue.dev, VS Code Copilot Chat, JetBrains AI Assistant, Claude Desktop, ChatGPT (remote URL), and Replit Agent (remote URL). Three install paths cover everything:
pip install patent-client-agents
from patent_client_agents.google_patents import GooglePatentsClient
async with GooglePatentsClient() as client:
patent = await client.get_patent_data("US10123456B2")
print(patent.title)
print(patent.abstract)
results = await client.search_patents(keywords=["machine learning neural network"])
for r in results.results:
print(f"{r.publication_number}: {r.title}")
The fastest path — nothing to install. Point any MCP-speaking client (Claude Code, OpenAI Codex CLI, Google Gemini CLI, Cursor, Windsurf, Cline, Zed, Continue.dev, VS Code Copilot Chat, JetBrains AI, Claude Desktop, ChatGPT Apps, Replit Agent, CoWork, …) at the public demo at mcp.patentclient.com:
{
"mcpServers": {
"patent-client-agents": {
"url": "https://mcp.patentclient.com/mcp"
}
}
}
Most clients also expose a "custom connector" / "add MCP server" UI that takes just the URL https://mcp.patentclient.com/mcp — no tokens to paste. On first connect you'll be sent to Google sign-in; approve and you're in. Any verified Google account works. Usage is rate-limited per account (100 MB/day, 20 MB/minute).
This is a public demo — don't send confidential material through it. See the Terms of Use.
111 patent + IP MCP tools are exposed by default. Credentialed families register when their environment variables are present, bringing the local/private surface up to 168 tools when every env-gated family is configured.
| Variable | Source | Required | How to get |
|---|---|---|---|
USPTO_ODP_API_KEY | USPTO ODP | Most USPTO patent tools | [developer.uspto.gov](https://developer.uspto.gov/) (free) |
USPTO_TSDR_API_KEY | USPTO TSDR | All TSDR trademark tools | [account.uspto.gov/api-manager/](https://account.uspto.gov/api-manager/) (free MyUSPTO account) |
EPO_OPS_API_KEY, EPO_OPS_API_SECRET | EPO OPS | All EPO tools | [developers.epo.org](https://developers.epo.org/) (free) |
JPO_API_USERNAME, JPO_API_PASSWORD | JPO | All JPO library + MCP tools (env-gated on the stdio server / plugin; not set on the hosted demo) | [j-platpat.inpit.go.jp](https://www.j-platpat.inpit.go.jp/) |
CANLII_API_KEY | CanLII | All CanLII library + MCP tools (env-gated on the stdio server / plugin; not set on the hosted demo) | [canlii.org/en/feedback/feedback.html](https://www.canlii.org/en/feedback/feedback.html) (free, by request) |
EUIPO_CLIENT_ID, EUIPO_CLIENT_SECRET | EUIPO | All EUIPO library + MCP tools (env-gated; not set on the hosted demo). Set EUIPO_ENV=sandbox to use the open sandbox environment instead of production. | [dev.euipo.europa.eu](https://dev.euipo.europa.eu/) (sandbox auto-approves; production requires ID-document review) |
USITC_EDIS_TOKEN | USITC EDIS | EDIS document/attachment downloads (also rejected for *public* docs without a token); investigation+document search itself works without one | [edis.usitc.gov](https://edis.usitc.gov) → API Token Generator (free, Login.gov account). JWT, ~2 wk lifetime |
USITC_DATAWEB_TOKEN | USITC DataWeb | run_dataweb_report only | [dataweb.usitc.gov](https://dataweb.usitc.gov) account page (free) |
PCA_WAF_TOKEN_PATH *or* PCA_WAF_TOKEN_JSON | USPTO TESS | Trademark search via TESS — bring-your-own WAF token *or* install [tmsearch] extra to mint via Playwright | See [USPTO Trademark Search docs](https://docs.patentclient.com/api/uspto-tmsearch/) |
No API key needed: Google Patents, USPTO Publications (PPUBS), USPTO Assignments, USPTO Trademark Assignments, MPEP, TMEP, CPC, WIPO Lex, Federal Circuit (CAFC), USITC HTS, USITC IDS, US Copyright Office.
/plugin marketplace add parkerhancock/patent-client-agents
/plugin install patent-client-agents@patent-client-agents
/reload-plugins
高质量的开源AI工作流工具
AI Skill Hub 为第三方内容聚合平台,本页面信息基于公开数据整理,不对工具功能和质量作任何法律背书。
建议在沙箱或测试环境中充分验证后,再部署至生产环境,并做好必要的安全评估。
✅ Apache 2.0 — 宽松开源协议,可商用,需保留版权声明和 NOTICE 文件,含专利授权条款。
总体来看,专利客户代理 是一款质量良好的Agent工作流,在同类工具中具备一定竞争力。AI Skill Hub 将持续追踪其更新动态,建议收藏备用,结合自身场景选择合适时机引入使用。
| 原始名称 | patent-client-agents |
| 原始描述 | 开源AI工作流:Intellectual property data tools for AI agents。⭐32 · Python |
| Topics | AI知识产权工作流 |
| GitHub | https://github.com/parkerhancock/patent-client-agents |
| License | Apache-2.0 |
| 语言 | Python |
收录时间:2026-06-06 · 更新时间:2026-06-08 · License:Apache-2.0 · AI Skill Hub 不对第三方内容的准确性作法律背书。
选择 Agent 类型,复制安装指令后粘贴到对应客户端