Awesome: change the play button to start Spotify instead of that crappy Music app. lowtechguys.com/musicdecoy/

Best commands/skills for local code reviews: ⭐️ /code-review built into Claude (use max effort) ⭐️ /review-pr from Claude's pr-review-toolkit plugin ⭐️ /thermo-nuclear-code-quality-review from Cursor

I've always had trouble deciding whether to wrap Markdown lines (at around 80 chars) or let them continue indefinitely. Semantic Line Breaks seem like a practical, diff-friendly alternative, esp. for AI-generated content. There's even a sembr-reformat skill. sembr.org

"Audited. Straight answer: your instinct is wrong here. This codebase is cool, not feverish—roughly 3/10 on the indirection scale where 10 is the cannon you quoted. The shape of the code is visible right through it." 😮‍💨

/code-review max 👌

Top Codex annoyance: /compact often discards too much context. Here, two skills found 5 refactor opportunities. Ran /compact before starting refactor 1) because it was approaching 100%. Codex does 1) well. /compact again. Codex no longer knows what tasks 2-5 were.

Although I don't have access to a Windows machine for debugging, I've found that Claude is really good at writing & running probe jobs in CI (windows-latest). You can get quite far with a few experiments like that.

was curious about Bun post-Anthropic acquisition and what it could tell us about the company's stewardship and intentions, and what it looks like when a key infrastructure project is increasingly maintained by AI. the numbers are interesting, with two big takeaways. redmonk.com/sogrady/2026...

This idea from @jessehanley.bsky.social resonated with me so strongly, I had to write about it. "Do the hardest thing." justinjackson.ca/hard-thing

One of the benefits of maintaining two DuckDB client libs is that I keep finding and fixing bugs in DuckDB itself github.com/duckdb/duckd...

Good morning! Starting the day by deleting 1500 LOC from duckdb-rs 💪 github.com/duckdb/duckd...

Using /grill-me to figure out how to classify my freelance work for the client's "new invoicing system"... 😑 github.com/mattpocock/s...

Go xhigh or go home

Took dynamic workflows for a spin. Took me a while to get the prompt right: disable-model-invocation prevents spawned agents from calling skills as a slash command (which is confusing/wrong). Happy with the result once everything is working, though workflows are a real token burner!

Claude starting the Docker daemon for debugging without asking me is both unsettling and cool (exactly what I would have done anyway).

In contrast, Claude Code's new /code-review is pretty good, even though there's a bug that causes findings to be reported in JSON.

Does the @coderabbitai.bsky.social CLI actually find anything useful on the free plan? Always reports "No findings" for me.

I use /improve-codebase-architecture and /prototype almost every day, and have created my own skills based on patterns from this repo (ADRs are a real game changer). Highly recommend. The 100k+ stars are absolutely well-deserved. github.com/mattpocock/s...

scc --avg-wage 150000 reports that 13k lines of Rust code amount to an "Estimated Cost to Develop" of $1M. Not unreasonable. However, 1500 lines of (AI-generated) Markdown will add another $200k. Pre-LLM cost estimates are (even more) fiction.

The Codex App has removed the context usage indicator from its UI. Apparently because auto-compaction is supposed to just work. Bad move in my opinion. github.com/openai/codex... Context pollution is still a thing: www.aihero.dev/what-is-the-...

I've contributed 5 PRs to this DuckDB release. Only one of them was reverted almost immediately. 😄🤙 github.com/duckdb/duckd...

Tokenneid (n., German) — envy directed at engineers with higher agent limits, typically experienced while staring at a "try again in 2h 14m" message.

Every time I've used LLMs to port libraries based on a test suite, I've discovered (and fixed) problems in the test suite. Funnily enough, this often makes the ported version more compliant than the original. github.com/amazon-ion/i...

Jarred rewrite Bun in Rust in a week with clankers. Say what you want about it, but that’s impressive. github.com/oven-sh/bun/...

I'm obsessed with all things John Ousterhout right now. If you don't know him yet, this is a great place to start. newsletter.pragmaticengineer.com/p/the-philos...

"No deepening candidates worth proposing. The crate has been through ~30 rounds of deliberate deepening (the git log reads like an architecture workshop)." Fair enough 😆 I pushed /improve-codebase-architecture pretty hard. 15 ADRs for a brand new Rust lib.

This must be one of the most useful skills for software dev out there. It's incredibly effective. www.youtube.com/watch?v=3MP8...

"So the phenomenon of engineers becoming thin wrappers around Claude Code is limited to the kind of engineers for whom this is an improvement in their work product." www.seangoedecke.com/ai-makes-wea...

. 🎞️ 𝗕𝗲𝘆𝗼𝗻𝗱 𝘁𝗵𝗲 𝗔𝗜 𝗛𝘆𝗽𝗲 𝟮𝟬𝟮𝟲: 𝗩𝗶𝗱𝗲𝗼 & 𝗦𝗹𝗶𝗱𝗲 𝗗𝗿𝗼𝗽 - Thanks to the OMR crew for the lightning-fast turnover www.youtube.com/watch?v=YNav... Slides on my Substack (free)

Higher usage limits for Claude and a compute deal with SpaceX https://www.anthropic.com/news/higher-limits-spacex (https://news.ycombinator.com/item?id=48037986)

Den besten deutschen Tech-Analysten mal live sehen ✅ Kein Talk war so gut (und so voll) wie der von @pip.net #OMR

I'm bullish on skills (unsurprisingly) and have started converting all of my commands. Just feels magical to be able to say "update Rust" or "why is the market down?" ✨

Claude dissing my portfolio 😆 "smart money doesn't want to own a 5-beta crypto-miner-pretending-to-be-an-AI-stock when MSFT could tank the sector with one cautious comment about 2027 capex" "For high-beta crap like your IREN/HUT, weekend gap risk is real (5x beta = 2% market gap = 10% on you)"

Parameter bindings just got easier in duckdb-rs - Slick new named_params! macro for mixed Rust types - HashMap support for dynamic params github.com/duckdb/duckd...

I wonder why GitHub has been getting so much hate lately (hard reload doesn't help)

Unintended consequence #417 of no-AI policies: Bun forking Zig and not contributing back x.com/bunjavascrip...

duckdb-go now has Typed() to force specific scalar types when binding query params (not tagged yet, just shipped it) github.com/duckdb/duckd...

"A common tragic figure today is the professional athlete who believes the show will go on forever and doesn’t prepare for the day they can’t do it anymore. We may be in the first generation of software engineers in the same position. If so, it’s probably a good idea to plan accordingly."

Freue mich schon auf die Talks von @pip.net und @steipete.me 🤙

❯ compare stock perf of HUT and MU and tell me which is the better bet, include recent news in the analysis I wrote a Yahoo Finance skill, and now Claude is giving me investment advice despite telling me otherwise ("Not financial advice").

Pi agent extension to tell the user to go to bed after midnight 😄👍 github.com/mitsuhiko/ag...

I like to read a good AI critique, but find it hard to take those seriously where the author lacks any practical experience ("I continue to write all of my software by hand") and only rehashes tired arguments: hallucinations, sociopaths, "they write bugs", etc. aphyr.com/posts/418-th...

Crazy what github.com/ranaroussi/y... does to scrape market data from Yahoo 😳

We are hosting the next edition of our community conference, DuckCon, in Amsterdam on June 24. The registration is already open and you can submit your talk proposals until next Wednesday. Looking forward to your submissions and seeing the flock in the summer!

I fixed mathias.blog/tools/whatzi..., and now the new Zig version shows up in all its glory. ⚡️ corsproxy.io has started to block free usage from production origins. As a workaround, I switched to codetabs.com (cool 90s-style website, btw).

Look who made it into the list of contributors again 👀

Published new versions of duckdb-rs and duckdb-go for yesterday's v1.5.2 release github.com/duckdb/duckd... github.com/duckdb/duckd...

How can GPT-5.4 (xhigh) be such a good model and yet produce the most useless commit messages imaginable? It's almost as if it only remembers the last thing it did and ignores the diff itself.

I turned this into a skill. Now I can simply ask ❯ /bear-notes what did i do this month for client X? ❯ /bear-notes what is Y? or just ❯ create a note about Z 😍

😅