Ratatosk
CNCF release intelligence: typed, quoted facts (security fixes, breaking changes, deprecations) from every graduated/incubating project's release notes, updated daily. check_stack compares running versions locally — versions never leave your process.
README
<div align="center">
<img src="docs/assets/ratatosk-hero.webp" width="280" alt="Ratatosk, a messenger squirrel with a scroll in its bag, waving hello">
ratatosk-mcp
Ratatosk reads CNCF release notes every hour. Your agents get the facts.
</div>
In Norse myth, Ratatosk is the squirrel that carries messages up and down the world tree. This one carries release intelligence. ratatosk.io watches 74+ CNCF projects and turns every release note into typed, entity-level facts: security fixes, breaking changes, removals, deprecations, changed defaults. Plain bug fixes and marketing copy are filtered out. What remains is what an operator acts on.
This MCP server hands those facts to your agent as tools.
What it feels like
You: "We run envoy v1.36.8 and istio 1.30.1. Anything we must do before upgrading?"
Your agent calls
check_stackand answers from facts: the CVEs fixed after your version, the APIs removed on your upgrade path, the defaults that changed. Each fact carries a verbatim quote from the release notes as evidence.
Tools
| Tool | What it does |
|---|---|
list_facts |
Incremental fact feed. Filter by project, type, severity; poll with the since cursor |
facts_by_entity |
Reverse index: every fact touching one identifier (CVE id, CRD, feature gate, flag, config field, dependency) |
list_projects |
The tracked-project roster — resolve slugs here first, never guess |
get_release |
One reviewed release: coverage, assessment, source, and all its facts. Omit version for the latest reviewed release of the project. facts: [] with coverage: full_reviewed means the release was read and is routine. include_raw adds the original note body (raw_notes) — automatic when the review is not the full story |
check_stack |
Takes the component versions you run, returns a briefing on your upgrade path: critical/high facts in full, one line each for the rest, the same advisory across release branches folded into one entry. detail: "full" for everything verbatim, target_version for one upgrade hop, severity_min to filter |
Your versions stay on your side
check_stack sends only project slugs to the server and compares version keys
locally, inside this process. What you run never reaches ratatosk.io. The
server publishes facts; your agent decides what applies. The version
normalizer is bundled (internal/version), so range comparison happens
client-side too.
Quick start
claude mcp add ratatosk -- docker run -i --rm ghcr.io/garlickim21/ratatosk-mcp:latest
No account, no API key. Running agents in Kubernetes, or using kagent? See the install guide below.
Documentation
| Install & usage — local stdio · in-cluster (Helm) · kagent | 한국어 · 日本語 |
| Helm chart — values, kagent toggle | |
| kagent example — manifests + ratatosk-agent | |
| Contributing · Security policy |
Upstream API
This server is a thin client over the public REST API. If you would rather
call it directly, GET /v1 on ratatosk.io describes itself. No API key; rate
limited at 60 requests per minute per IP.
Data & terms
The data is served by ratatosk.io free of charge (subject to change, with
advance notice) under its terms of service.
Analyses are AI-generated reference information with no warranty — check the
original release notes before acting, especially when an agent acts on your
behalf. Original notes belong to their respective projects; responses that
carry a full note include an attribution notice (raw_notes_notice).
License
The code in this repository is licensed under Apache-2.0.
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。