Managed AI Agents for Data Analysis https://teapawt.org/t.co/WmWnVz4dQy
cloud
Joined September 2013
- Tweets889
- Following282
- Followers212
- Likes108
An AI agent shouldn't be twelve YAML keys you hold in your head. Our visual editor draws it: the agent in the middle, its model, MCP servers, rules, skills around it. Canvas, spec and Python are one definition.
Try it, no sign-in โ datalayer.ai
#AIAgents #MCP
๐ Agent specs are getting nicer: Oracle's Open Agent Spec, Pydantic AI, runtime-enforced AgentSpec.
Datalayer Agentspecs now makes protocols specs too: MCP, A2A, PAP, PACTโฆ declared per agent, checked, compiled to Python + TypeScript
๐งฉ github.com/datalayer/agentspโฆ
#AIAgents
๐ Personal Agent Protocol 0.11.0: the first complete Python ๐ + TypeScript ๐ฆ SDKs, both roles, an example for every feature
๐งญ personal-agent-protocol.dataโฆ
Now integrating it in Datalayer
๐ personal-agent-protocol.dataโฆ
๐ Datalayer's Poppy SDKs (Python ๐ + TypeScript ๐ฆ) are listed on poppy.md!
๐ Must-read: Poppy vs MCP, A2A, UCP, ACP, TAP, PACT. Poppy answers the consent question ๐
Kudos Fabricio Telles ๐
poppy.md/compare
๐ช ๐งย Jupyter MCP Server 2.2.4 ships even more fixes and documentation for users. The release notes are available on lnkd.in/ewgSbUNw
Community support stays strong ๐ช, with 2 new contributors หห๐ขึดเป๐โงห.๐โ
๐ Our Personal Agent Protocol SDKs (Python ๐ + TypeScript ๐ฆ) are on awesome-poppy!
Draft 0.1 today; sign-in, conversations and Operations next ๐
Thanks Javier Chulvi Bernad for the list ๐
github.com/JaviChulvi/awesomโฆ
๐ PAP SDK v0.3.2 is live for Python + TypeScript!
PAP is an open protocol for every company and every personal agent. We โค๏ธ
protocols and want to build with the community.
๐ธ personalagentprotocol.org/\
โจ Listed on github.com/JaviChulvi/awesomโฆ
#poppy
2/3 ๐ Marimo, now a toolset of the MCP server: mcp.datalayer.run/mcp?marimo
Cell a: x = 1. Cell b: print(x + 1) โ 2. Change a to x = 10 โ the answer carries b re-run, 11 printed. The agent never tracks what depended on what.
As we ran it: datalayer.ai/docs/mcp/toolseโฆ #Marimo
3/3 ๐ค Agent Workers run for hours in your notebook, beside your team. ๐ Loop: a Notion-like write-up over the live notebook โ no account: datalayer.ai/try/loop. ๐ Benchmarks reopen a failed task where it ran.
Which would you use first? ๐
๐ช Jupyter MCP Server 2.2.3 is outโpowered again by community contributions. Thanks @kta1kri (safer HTTP), @sachhg (credential isolation), @nickilicious (smaller image responses) & @pma1999 (tool routing)! ๐ค
github.com/datalayer/jupyterโฆ
โจ An agent that answers with an interface, not a wall of text.
๐
A form that sends your input back to the agent;
๐ Jupyter output as an A2UI surface, buttons that send your click to the code.
๐ฌ 90 s, ๐ open source โ github.com/datalayer/agent-rโฆ
#A2UI #Jupyter
Before, changing someone else's tool meant registering one with the same name and hoping yours loaded second. It worked on one machine and did nothing on the next.
Now you declare it โ ("launch_sandbox", ToolExtension(wrap=โฆ)) โ and get one tool, applied in a fixed order.
It runs on Reactor, our open-source plugin platform (VS Codeโstyle contribution points, Python and TypeScript), through reactor-mcp-server: an MCP foundation any server can build on.
Release notes: jupyter-mcp-server.datalayerโฆ
Repo: github.com/datalayer/jupyterโฆ
๐ ๐๐ฉ๐ค๐ฅ ๐ฅ๐๐จ๐จ๐๐ฃ๐ ๐
๐ช๐ฅ๐ฎ๐ฉ๐๐ง ๐ฃ๐ค๐ฉ๐๐๐ค๐ค๐ ๐จ ๐๐ง๐ค๐ช๐ฃ๐.
Handoffs lose state, files, outputs and context.
๐ช Datalayer gives teams and AI agents one live Jupyter notebook.
Same cells. Same state. Same results.
datalayer.ai
#Jupyter #Collaboration
๐ ๐๐๐ฝ๐๐๐ฒ๐ฟ ๐ณ๐ผ๐ฟ ๐๐ ๐๐ด๐ฒ๐ป๐๐
Our new Datalayer homepage is live.
Give your agent persistent notebooks, scalable compute & connected dataโor try an Agent built for Jupyter directly.
The work keeps running when you disconnect.
๐ datalayer.ai
๐ Jupyter MCP Server 2 is here.
Now on MCP Python SDK 2.
No breaking change for users.
What it unlocks:
โก stateless scale
๐ stronger auth
๐ richer agent flows
๐งฉ async + long-running tasks
๐๐๐ฝ๐๐๐ฒ๐ฟ + ๐ ๐๐ฃ ๐ถ๐ ๐ท๐๐๐ ๐ด๐ฒ๐๐๐ถ๐ป๐ด ๐๐๐ฎ๐ฟ๐๐ฒ๐ฑ.
๐ ๐ง๐ต๐ฒ ๐ต๐ผ๐๐๐ฒ๐ฑ ๐๐๐ฝ๐๐๐ฒ๐ฟ ๐ ๐๐ฃ ๐ฆ๐ฒ๐ฟ๐๐ฒ๐ฟ ๐ถ๐ ๐น๐ถ๐๐ฒ.
๐ One MCP endpoint gives your AI agent:
๐ Persistent notebooks
โณ Durable execution
โก CPU/GPU compute
๐๏ธ Direct data access
๐ค Human + agent collaboration
๐ datalayer.ai/docs/mcp
