Lightnews — Scholar-powered news

Reposted by Sebastian Korfmann

goose @opensource.block.xyz · 6h

Stop repeating yourself: goose can build tools from your work history!

🔍 Analyze session data
📖 Build recipe automation
📈 Level up productivity
⏰ Save your future self time!

1 3 5

Sebastian Korfmann @skorfmann.com · 3d

what kind of queries / access patterns?

1

Sebastian Korfmann @skorfmann.com · 3d

just stuff it into parquet files and dump it into a bucket 😅

1 1

Sebastian Korfmann @skorfmann.com · 3d

There was a talk about the Agent Client Protocol on our last agentic coding meetup here in Hamburg - glad so see that the protocol gets broader adoption zed.dev/blog/jetbrai...

ACP Brings JetBrains on Board - Zed Blog

From the Zed Blog: The Agent Client Protocol reaches a major milestone: JetBrains is committing to bringing ACP support to their entire ecosystem.

zed.dev

Reposted by Sebastian Korfmann

Taylor Mullen @ntaylormullen.bsky.social · 4d

We've released Gemini CLI extensions for querying PostgreSQL and MySQL with natural language.

If you're interested, take them for a spin! Oh and we'd appreciate any feedback you'd be willing to share 👇

github.com/google-gemin...

#GeminiCLI #Postgres #MySQL #OSS

4 6

Sebastian Korfmann @skorfmann.com · 4d

mostly manual context building (with help of Claude Code or claude.ai) and then keeping track of changes / decisions in a file plus maintaining some sort of quick reference index file about the project (which component does what and where). Those frameworks felt a bit too strict for my cases

Reposted by Sebastian Korfmann

Tobias Müller @tobilg.com · 4d

Yup. Claude coded a complete DuckDB Extension for me during my lunchbreak with the help of three MCP servers and a handful prompts… Crazy.

1 2 5

Sebastian Korfmann @skorfmann.com · 4d

At this point, ignoring agentic coding tools is like refusing to use Git because you prefer manually tracking file versions in folders named 'final_v2_ACTUAL_final'

3 1 10

Sebastian Korfmann @skorfmann.com · 4d

Yes! Also, research tasks in chatgpt or Claude.ai to generate use case specific docs

1

Reposted by Sebastian Korfmann

Tobias Müller @tobilg.com · 4d

Created a DNS extension for @duckdb.org today:

github.com/tobilg/duckd...

Hopefully it will be available via the Community Extensions soon!

GitHub - tobilg/duckdb-dns: DNS (Reverse) Lookup Extension for DuckDB

DNS (Reverse) Lookup Extension for DuckDB. Contribute to tobilg/duckdb-dns development by creating an account on GitHub.

github.com

1 4 16

Sebastian Korfmann @skorfmann.com · 4d

Prototyping with Claude Sonnet 4.5 is incredible 🧑‍🎨

1

Sebastian Korfmann @skorfmann.com · 5d

Compacting a Claude session early can be like feeding a sourdough starter - discard the old stuff, keep the good parts, watch it rise again 🍞✨

1 1 4

Sebastian Korfmann @skorfmann.com · 6d

Clauding again... 🔥

1 4

Sebastian Korfmann @skorfmann.com · 6d

The Claude Agent SDK is pretty cool 😎 docs.claude.com/en/api/agent...

Agent SDK overview - Claude Docs

Build custom AI agents with the Claude Agent SDK

docs.claude.com

1 8

Sebastian Korfmann @skorfmann.com · 8d

Was just thinking today, that I need something which prompts me every now and then to quickly write (dictate) some summary / notes 😅

1

Sebastian Korfmann @skorfmann.com · 8d

I'm a markdown junkie now thanks to Claude Code - been doing tons of code analysis lately. But viewing them in the terminal sucks - luckily @thieso2.bsky.social built www.npmjs.com/package/more... - a handy tool to serve your directory and navigate/search md files - with Mermaid rendering! 🎉🚀

www.npmjs.com

1 5

Reposted by Sebastian Korfmann

Taylor Mullen @ntaylormullen.bsky.social · 13d

And that's a wrap on Gemini CLI's 3-day drop! What started as 3 days got a little bigger 😉

Huge thanks to everyone who followed along!

Full recap of what we launched this week 🧵👇

#GeminiCLI #AI #DevTools #OSS #3DayDrop

1 1 3

Reposted by Sebastian Korfmann

Tobias Müller @tobilg.com · 14d

@cloudflare.social's data platform was launched today:

blog.cloudflare.com/cloudflare-d...

It consists of the following parts:

- Pipelines: Receive & transform events
- R2 Data Catalog: Manages the Iceberg metadata & table maintenance
- R2 SQL: In-house distributed SQL engine based on Data Fusion