发布 HN：Strata（YC X25）——一个MCP服务器，专为AI处理数千个工具而设计
大家好！我们是 Klavis AI（[https://www.klavis.ai](https://www.klavis.ai)），今天我们推出了 Strata，一个开源的 MCP 服务器，帮助 AI 代理使用成千上万的 API 工具而不至于感到不知所措。Strata 不会一次性展示所有可用工具，而是根据 AI 实际需要逐步揭示。
作为谷歌 Gemini 工具使用团队的前高级软件工程师，我亲眼目睹了 AI 在使用工具时的困难。如果你曾经构建过 AI 代理，可能也遇到过同样的问题：（1）AI 代理在数百个选项中选择合适的 API 变得困难；（2）工具描述和信息消耗大量的令牌预算；（3）大多数服务器将工具数量限制在 40~50 个，以避免这些问题，从而限制了你的构建能力。
Strata 的工作方式更像人类。它引导 AI 代理发现相关类别，然后列出这些类别中的可用操作。它依赖于大语言模型的推理，逐步深入找到所需的确切工具。以下是一些示例：
*Github 查询：“在我们的主仓库中找到我的过期拉取请求”*
Strata：AI 模型识别 GitHub → 显示类别（仓库、问题、拉取请求、操作）→ AI 选择拉取请求 → 显示拉取请求特定操作 → AI 选择 list_pull_requests → 显示 list_pull_requests 详情 → 使用正确的参数执行 list_pull_requests。
*Jira 查询：“在‘MOBILE’项目中创建一个关于应用程序启动时崩溃的错误票据。”*
Strata：AI 识别 Jira → 显示类别（项目、问题、冲刺）→ AI 选择问题 → 显示操作（create_issue, get_issue）→ AI 选择 create_issue → 显示 create_issue 详情 → 使用正确的参数执行。
*Slack 查询：“在 #announcements 频道发布一条消息，通知下周五将发放奖金。”*
Strata：AI 识别 Slack → 显示类别（频道、消息、用户）→ AI 选择消息 → 显示操作（send_message, schedule_message）→ AI 选择 send_message → 显示 send_message 详情 → 使用正确的参数执行。
这种逐步的方法带来了巨大的优势：深度。虽然大多数集成只提供少量高层次的工具，但 Strata 可以为单个应用（如 GitHub、Jira 等）提供数百个细粒度的功能。你的 AI 代理终于可以访问真实工作流程所需的深层次、具体的功能，而不会在选项的海洋中迷失。
在后台，Strata 管理身份验证令牌，并包括一个内置搜索工具，供代理在遇到困难时查阅文档。
在 MCPMark（[https://mcpmark.ai/leaderboard/mcp](https://mcpmark.ai/leaderboard/mcp)）上，Strata 的 pass@1 率比官方 GitHub 服务器高出 15.2%，比官方 Notion 服务器高出 13.4%。在人工评估测试中，它在复杂的真实多应用工作流程中达到了 83% 以上的准确率。
这里有一个快速演示，展示了 Strata 如何在多个应用中导航复杂工作流程，并在每一步自动选择正确的工具：[https://www.youtube.com/watch?v=N00cY9Ov_fM](https://www.youtube.com/watch?v=N00cY9Ov_fM)。
你可以将任何外部 MCP 服务器连接到 Strata，我们也提供了一个开源版本：[https://github.com/Klavis-AI/klavis](https://github.com/Klavis-AI/klavis)。
如果你需要更多功能的团队或生产使用，请访问我们的网站：[https://www.klavis.ai](https://www.klavis.ai)。只需一键即可将 Strata 添加到 Cursor、VS Code 或任何兼容 MCP 的应用程序中。你还可以使用我们的 API，轻松将 Strata 集成到你的 AI 应用中。
期待你的反馈。感谢阅读！
查看原文
Hey HN! We are Klavis AI (<a href="https://www.klavis.ai/">https://www.klavis.ai/</a>) and we're launching Strata, one open-source MCP server that helps AI agents use thousands of API tools without getting overwhelmed. Instead of showing all available tools at once, Strata reveals them step-by-step based on what the AI actually needs.<p>As a former Senior SWE on Google Gemini 's tool use team, I saw firsthand how AI would struggle with tools. If you've built AI agents, you've likely hit the same walls: (1) AI agents struggle to pick the right API from hundreds of options. (2) Tool descriptions and info consume massive token budgets. (3) Most servers cap at 40~50 tools to avoid these problems, limiting what you can build.<p>Instead of flooding the AI with everything upfront, Strata works like a human would. It guides the AI agents to discover relevant categories, then lists available actions in those categories. It relies on LLMs’ reasoning to drill down progressively to find the exact tool needed. Here are some examples:<p><i>Github query: "Find my stale pull requests in our main repo"</i><p>Strata: AI model identifies GitHub → Shows categories (Repos, Issues, PRs, Actions) → AI selects PRs → Shows PR-specific actions -> AI selects list_pull_requests → Shows list_pull_requests details -> Executes list_pull_requests with the right parameters.<p><i>Jira query: "Create a bug ticket in the 'MOBILE' project about the app crashing on startup."</i><p>Strata: AI identifies Jira → Shows categories (Projects, Issues, Sprints) → AI selects Issues → Shows actions (create_issue, get_issue) → AI selects create_issue → Shows create_issue details → Executes with correct parameters.|<p><i>Slack query: "Post a message in the #announcements channel that bonus will be paid out next Friday."</i><p>Strata: AI identifies Slack → Shows categories (Channels, Messages, Users) → AI selects Messages → Shows actions (send_message, schedule_message) → AI selects send_message → Shows send_message details → Executes with correct parameters.<p>This progressive approach unlocks a huge advantage: depth. While most integrations offer a handful of high-level tools, Strata can expose hundreds of granular features for a single app like GitHub, Jira, etc. Your AI agent can finally access the deep, specific features that real workflows require, without getting lost in a sea of options.<p>Under the hood, Strata manages authentication tokens and includes a built-in search tool for the agent to dig into documentation if it gets stuck.<p>On the MCPMark <a href="https://mcpmark.ai/leaderboard/mcp" rel="nofollow">https://mcpmark.ai/leaderboard/mcp</a>, Strata achieves +15.2% higher pass@1 rate vs the official GitHub server and +13.4% higher pass@1 rate vs the official Notion server. In human eval tests, it hits 83%+ accuracy on complex, real-world multi-app workflows.<p>Here is a quick demo to watch Strata navigate a complex workflow with multiple apps, automatically selecting the right tools at each step: <a href="https://www.youtube.com/watch?v=N00cY9Ov_fM" rel="nofollow">https://www.youtube.com/watch?v=N00cY9Ov_fM</a>.<p>You can connect to any external MCP Server into Strata, and we have an open source version for it: <a href="https://github.com/Klavis-AI/klavis" rel="nofollow">https://github.com/Klavis-AI/klavis</a>.<p>For team or production use with more features, visit our website: <a href="https://www.klavis.ai">https://www.klavis.ai</a>. Add Strata to Cursor, VS Code or any MCP-compatible application with one click. You can also use our API to easily plug in Strata to your AI application.<p>We look forward to your comments. Thanks for reading!