Laxis
io.github.laxis-tech由 GitHub 账号 laxis-tech 发布v0.1.0更新于 Oct 24, 2026
Search and read your Laxis meeting transcripts, AI summaries, and participants from Claude.
概览
托管的 Laxis MCP 服务器,让助手搜索并读取你的会议转录、AI 摘要和参会人信息。
- 可以做什么
- 浏览近期会议,含标题、日期、时长和简短摘要(仅元数据)。对转录内容做语义搜索,返回最相关的片段。读取单场会议的参会人、AI 摘要要点和元数据。获取完整的分页转录文本,带时间戳和说话人标注。
- 适用场景
- 适合让助手回忆过往通话中说过或约定的事项、快速补看一周的会议、在转录中定位某个具体时刻,或借助早前上下文为下一场通话做准备。
- 运行要求
- 远程 Streamable HTTP 端点,无需本地安装。需要 Laxis 账户,首次连接时通过 OAuth 登录(OAuth 2.1 + PKCE)。仅对无法交互登录的客户端提供个人 API 密钥,以 Authorization Bearer 头发送。
安装
在 SourceWeft 中
- 打开 控制台中的 Laxis,将其添加到工作区。
- 连接 所需的 凭据。它们会被私密存储,绝不会在公开页面上展示。
- 为需要使用其工具的对话启用该服务。
Web executable,通过 Streamable HTTP。 远程服务在工作区中配置后即可从网页运行时运行。
交给智能体安装
把这段提示词粘贴到 Claude Code、Codex、Cursor 等智能体,或 SourceWeft 对话中。智能体会读取该服务器的安装指南,先向你展示发布者、端点和登录方式,经你同意后再把它添加到你的客户端。密钥不会写进任何命令。
阅读 https://sourceweft.com/mcp/io.github.laxis-tech%2Flaxis-mcp/install.md,并按其中说明把这个 MCP 服务器接入我的客户端。修改任何配置之前,先告诉我它的发布者、端点和登录方式,等我确认。不要把 API 密钥写进命令或我会提交的文件;问我密钥应从哪里获取。其他 MCP 客户端
把它添加到你客户端的 mcpServers 配置中。 客户端会在首次连接时完成 OAuth 登录。
{
"mcpServers": {
"laxis": {
"type": "http",
"url": "https://app.laxis.tech/mcp"
}
}
}README
Laxis MCP Server
[Add to Cursor] [Install in VS Code] [License: MIT]
The Laxis MCP Server connects Laxis — the AI meeting note taker and voice dictation tool for professionals — to Claude and other AI tools through the Model Context Protocol. Your assistant can search and read your meeting transcripts, summaries, and participants directly in chat — turning every conversation you've recorded into context your AI can reason over.
Remote, hosted server — nothing to install or run. Laxis hosts the server for you. You add one URL to your AI client and sign in with your Laxis account — no local binary, Docker image, or
npmpackage, and no API key to copy or paste.
What is Laxis?
Laxis is the next-generation AI meeting note taker and voice dictation tool built for professionals. It records, transcribes, and summarizes any conversation directly from your computer — no "bot" ever joins the call — and works seamlessly with Zoom, Google Meet, Microsoft Teams, Webex, Slack Huddles, and more.
- 🎥 Botless meeting recording & transcription — record and instantly transcribe any meeting straight from your computer for a private, professional, and seamless experience. High-accuracy AI handles multiple speakers and technical jargon with ease.
- 🎙️ AI dictation — voice to text, anywhere — speak instead of type. Laxis turns your voice into clean, perfectly formatted text up to 3× faster than typing, automatically removing filler words ("um," "uh") and adding punctuation. Perfect for drafting emails, notes, and messages by voice.
- 🌐 Real-time translation — Laxis transcribes and translates conversations live, so you can follow and capture cross-language discussions as they happen.
- ✍️ Personalized AI summaries — don't just get a transcript. Laxis combines your own real-time notes with the meeting transcript to generate insights that reflect your specific context and priorities.
- 📊 50+ professional report templates — turn hours of conversation into structured documentation in seconds, including Sales Discovery, Project Updates, User Interviews, and Candidate Screenings.
- ✅ Automated action items — Laxis automatically identifies and extracts tasks, ensuring nothing falls through the cracks.
- 💬 Real-time AI copilot — chat with Laxis before, during, and after meetings to get real-time insight.
- 🔄 Seamless CRM & tool integration — automatically sync meeting insights, action items, and customer data directly into Salesforce, HubSpot, Jira, Slack, and your favorite productivity tools.
- 🔒 Enterprise-grade security — your data is encrypted and private, protected to the highest industry standards.
Sales teams use Laxis to update their CRM instantly and never miss a customer requirement; product managers turn user interviews into actionable PRDs; consultants deliver standardized client reports; recruiters capture every detail of candidate interviews. This MCP server brings all of that recorded knowledge to your AI assistant.
Use cases
- Recall decisions and commitments — "What did we agree on pricing in my last call with Acme?"
- Catch up fast — "Summarize my meetings from this week and list the action items."
- Find the exact moment — "Find where the customer mentioned their renewal date."
- Prep for the next call — "Who attended the Q3 planning meeting and what did we promise to follow up on?"
Tools
All tools are read-only. The server returns raw snippets from your meetings; your AI client generates the answer. It never writes to your account, makes LLM calls of its own, or touches data that isn't yours.
How you connect
You only need a Laxis account. Add the endpoint to your AI client and it will open your browser to sign in with Laxis (OAuth 2.1
- PKCE) the first time it connects — approve once and you're done. Access tokens refresh automatically, and you can revoke a client's access anytime from Laxis → Settings → Claude (MCP).
The connection endpoint is always:
We no longer recommend personal API keys. Signing in with OAuth is more secure (no long-lived secret to copy, store, or leak) and just as quick. API keys remain available only for tools that can't sign in interactively — see Connecting without a browser.
Install
Pick your client. In every case you connect with the URL above and sign in with Laxis when prompted — there's no key to paste.
Claude Desktop & claude.ai (web) — custom connector (recommended)
In Claude Desktop or on claude.ai go to Settings → Connectors → Add custom connector, paste the URL below, and complete the "sign in with Laxis" flow.
Claude Code
Then run /mcp in Claude Code and choose Authenticate to sign in with
Laxis. Ask something like "Use Laxis to summarize my last meeting."
Cursor
The one-click button adds the server; open Cursor → Settings → MCP and click
Sign in (or Needs login) next to Laxis to authorize. Or configure it
manually in ~/.cursor/mcp.json:
VS Code
Requires VS Code 1.101+. After installing, VS Code prompts you to sign in with
Laxis on first use. To configure manually, add a .vscode/mcp.json:
Windsurf
Add to ~/.codeium/windsurf/mcp_config.json, then authorize when prompted:
Other clients
Any client that speaks Streamable HTTP to a remote MCP server and supports the MCP OAuth flow works. Point it at:
- URL:
https://app.laxis.tech/mcp - Transport: Streamable HTTP
- Auth: OAuth 2.1 (the client discovers the sign-in flow automatically)
Connecting without a browser (API key)
Not recommended. Use this only for tools that can't complete an interactive OAuth sign-in — headless scripts, CI, or automations such as Zapier. A personal API key is a long-lived secret: anyone who has it can read your meetings until you revoke it.
-
In Laxis, go to Settings → Claude (MCP) → Advanced and generate a personal API key (
laxis_…). -
Send it as a bearer token on every request:
For a desktop client that lacks native OAuth, you can bridge with
mcp-remote and the header:
Treat the key like a password. Regenerating it immediately invalidates the previous one (and any automations that share it), so update those afterward.
How it works
The server is a thin, retrieval-only proxy in front of the Laxis API. It validates your OAuth access token, exchanges it for a short-lived, user-scoped upstream token (it never forwards your client's token to the API), fetches the requested meeting data, and returns raw snippets. Your AI client does all the reasoning and answer-writing. You only ever see your own meetings, and access stops the moment you revoke the connection (or, in API-key mode, the key).
Roadmap
- Listing in Anthropic's connector directory and on public MCP registries, for one-click discovery.
Support
- Questions / bugs: open an issue.
- Product help: laxis.com / in-app support.
- Security: see SECURITY.md — please report privately, not via public issues.
License
MIT © Laxis
来源:README.md,提交 b8ebbce
工具
0版本历史
1- v0.1.0最新Sep 16, 2026
评分
已满足适用检查项的 85%
该评分衡量此收录条目的完整程度,依据注册中心条目、端点与公开来源计算得出。它不是对服务器本身的评价,也不表示服务器安全或当前可用:可连通性是带日期的检查结果,评分也不包含安装量或用户评价。
- 近 30 天内可连通必需25 分
- 已声明认证方式必需10 分
- 已列出工具15 分
请让服务器在无凭据时响应 tools/list,或发布可在沙箱中无凭据启动的软件包。
- 描述必需15 分
- 图标10 分
- 文档15 分
- 许可证10 分