build an intelligent query layer with MCP in #elasticsearch — from search to agent.here's a preview how ES|QL queries turn into tools that get aggregated into agents — all served and built right in kibana. and you can try this out today

RT @DanielGri: Here is the recording of Tech Soirée Vienna - feat. @patrickhrmn @blwiertz @techeurope_ @getsentry @Sierralog @xeraa @moritz….

Join millions who have switched to Grok.

RT @prototype_cap: We had two awesome firesides!. First was on how to sell & distribute dev products in the AI-age, moderated by @Sierralog….

vienna is buzzing around AI. panel night: sell & distribute dev products in the age of AI | how to vibe code to a billion dollars

failure catalog of AI apps — not sure I've seen it spelt out that way before.not sure what to think of the product behind it but the list and ways to approach the problem look pretty useful:

RT @context_sdk: We’re super excited to co-host Tech Soirée Vienna: Infra & Dev in the Decade of AI alongside @FortinoCapital, Prototype by….

positive tech surprise of the day: @Ubiquiti .my family‘s router died a month ago — locked in a small, hot, and dusty cupboard for years.not only did the weekly cloud backup work and was restorable with all connected devices and settings; it even worked for the upgrade from the

how to get there is a little hidden right now but it's all public: on elasticsearch serverless it's already available (in other deployment options soon). you need to enable a couple of feature flags for it and then you're good to go:.POST

or you can take it to any MCP client:.{. "mcpServers": {. "elastic": {. "command": "npx",. "args": [. "mcp-remote",. " https://. /api/chat/mcp",. "--header",. "Authorization:${AUTH_HEADER}". ],. "env": {. "AUTH_HEADER":

to make trying this out super simple, there's a conversation interface right in kibana (on elastic cloud already with a built-in LLM, so you can just get going)

those tools can then be made available in different agents, so you can aggregate and connect your tools with A2A

the basic idea is that any ES|QL queries (with JOINs, hybrid search, aggregations, geo queries,. ) can turn into custom MCP tools. we provide a list of default ones to get mappings, get documents by ID, run a general query,. but the really interesting ones you'll have to write

history:.v0: we started by exposing elasticsearch's _search API through a standalone MCP — great to get something going.v0.5: adding search templates (where you can fill in placeholders) gave the concept quite a bit more power.v1: brand new preview directly in kibana.

MCP meets #elasticsearch — @mcpuse x @elastic: we had a great hack night in SF with projects around finding and using the right MCP tools. folks built:.* what tools do I need to prepare a heist.* find and download youtube videos (if your LLM thinks it‘s legal).* plan your

RT @npmzula: People might not realise, but elasticsearch is crazy good!.

and you got your RSS feed (that could also be turned into emails etc with some automation like IFTTT) 3/3.

1. from the page like or 2. find the underlying file on github (luckily everything is on github) — or 3. switch to "commits" and add ".atom" to the end of the URL —.

TIL how to.* subscribe to our release notes or .* the @elastic integrations (which are independent of stack releases).since two people asked a very similar question today. it's actually 3 simple steps 1/3

RT @pederzh: 🚀 tomorrow, come and hack with us to build the MCP discovery system using @elastic and @mcpuse!.the challenge is to build a be….