Lightnews — Scholar-powered news

Reposted by Hans L’Hoest

Meredith Whittaker @meredithmeredith.bsky.social · 7h

We're so relieved to see Germany reaffirm its opposition to the dangerous Chat Control proposal--the one that would mandate mass scanning of communications. Germany's long been a solid champion of privacy, and the news that it was considering backing mass surveillance was alarming. 1/

7 170 810

Hans L’Hoest @hanlho.com · 9h

I second this. Is what I did.

1

Hans L’Hoest @hanlho.com · 9h

For me, DORA’s annual research is always worth paying attention to, most importantly because of its scientific grounding and its practical insights that keep on evolving with our field.

DORA Report 2025 Infographic:
dora.dev/research/202...
Full report:
dora.dev/research/202...

dora.dev

Hans L’Hoest @hanlho.com · 1d

Ha! That’s when you bring up Yagni again 😄

1

Hans L’Hoest @hanlho.com · 2d

Instead adopt APYIAC : avoid painting yourself in a corner. That’s what I used to throw in the mix whenever someone brings up yagni in the wrong context.

Hans L’Hoest @hanlho.com · 3d

My main decision influences: cost control, flexibility and avoiding vendor lock-in (Claude Code's past month changes have this effect).

After spending a few months in the CLI I want to look again at working more in an IDE.

Note: I do not subscribe to Max plans, I combine multiple lower cost ones.

Hans L’Hoest @hanlho.com · 3d

I dropped Claude Code and did not subscribe to Codex.

If this turns out to be a bad idea, I can always try again.

1

Hans L’Hoest @hanlho.com · 3d

After wrapping my head around the constant changes in LLM subscriptions and performance, here’s my new coding stack for October:

- Warp Pro: my go-to agentic CLI, supports most top end LLMs.
- Zed Pro (using $20 trial)
- OpenCode for agentic CLI dev (local + cloud models, easy swapping, no lock-in)

1

Reposted by Hans L’Hoest

Meredith Whittaker @meredithmeredith.bsky.social · 3d

📣 Germany's close to reversing its opposition to mass surveillance & private message scanning, & backing the Chat Control bill. This could end private comms-& Signal-in the EU.

Time's short and they're counting on obscurity: please let German politicians know how horrifying their reversal would be.

Signal @signal.org · 6d

We are alarmed by reports that Germany is on the verge of a catastrophic about-face, reversing its longstanding and principled opposition to the EU’s Chat Control proposal which, if passed, could spell the end of the right to privacy in Europe. signal.org/blog/pdfs/ge...

signal.org

30 1.6K 2.3K

Reposted by Hans L’Hoest

Signal @signal.org · 6d

We are alarmed by reports that Germany is on the verge of a catastrophic about-face, reversing its longstanding and principled opposition to the EU’s Chat Control proposal which, if passed, could spell the end of the right to privacy in Europe. signal.org/blog/pdfs/ge...

signal.org

41 2.4K 3.9K

Hans L’Hoest @hanlho.com · 9d

Claude Code 2.0.0: `/rewind` killer feature

Anthropic's models are among the best although, as in the past weeks, wonky at times. But with their CLI they currently keep on having an edge on the competition.

(Disclaimer: I have not tried all of them but I know Codex and Gemini lag.)

1

Hans L’Hoest @hanlho.com · 9d

To reduce cognitive load: bounded contexts and clear contracts between them.

Hans L’Hoest @hanlho.com · 10d

Turns out Claude Code has a general setting that allows you to configure whether to include the co-authored-by Claude byline in git commits and pull requests! 🤦‍♂️ The default value of this setting is true. See settings.json and set the value to false. docs.claude.com/en/docs/clau...

Claude Code settings - Claude Docs

Configure Claude Code with global and project-level settings, and environment variables.

docs.claude.com

Hans L’Hoest @hanlho.com · 10d

I do not know if you are interested in trying out other LLM agents but when it comes to scripting I have had good results with Warp. It was able to figure out issues in scripts which I kept going around in circles with with Claude Code. (Comes with some monthly free credits.)

1

Reposted by Hans L’Hoest

Meredith Whittaker @meredithmeredith.bsky.social · Sep 9

📣 NEW -- In The Economist, discussing the privacy perils of AI agents and what AI companies and operating systems need to do--NOW--to protect Signal and much else!

www.economist.com/by-invitatio...

An illustration of me, and the headline: "AI agents are coming for your privacy, warns Meredith Whittaker
The Signal Foundation’s president worries they will also blunt competition and undermine cyber-security"

To put it bluntly, the path currently being taken towards agentic AI leads to an elimination of privacy and security at the application layer. It will not be possible for apps like Signal—the messaging app whose foundation I run—to continue to provide strong privacy guarantees, built on robust and openly validated encryption, if device-makers and OS developers insist on puncturing the metaphoric blood-brain barrier between apps and the OS. Feeding your sensitive Signal messages into an undifferentiated data slurry connected to cloud servers in service of their AI-agent aspirations is a dangerous abdication of responsibility.

Happily, it’s not too late. There is much that can still be done, particularly when it comes to protecting the sanctity of private data. What’s needed is a fundamental shift in how we approach the development and deployment of AI agents. First, privacy must be the default, and control must remain in the hands of application developers exercising agency on behalf of their users. Developers need the ability to designate applications as “sensitive” and mark them as off-limits to agents, at the OS level and otherwise. This cannot be a convoluted workaround buried in settings; it must be a straightforward, well-documented mechanism (similar to Global Privacy Control) that blocks an agent from accessing our data or taking actions within an app.

Second, radical transparency must be the norm. Vague assurances and marketing-speak are no longer acceptable. OS vendors have an obligation to be clear and precise about their architecture and what data their AI agents are accessing, how it is being used and the measures in place to protect it.

11 280 880

Hans L’Hoest @hanlho.com · Sep 2

Now you can ask Claude Code: `Can you propose a rewrite of ADR xxx in the form of Y-statement?` and see what it comes up with and if it suits your project, slide deck or t-shirt better. It's actually quite accurate.

1

Hans L’Hoest @hanlho.com · Sep 2

When it comes to exploring different decision recording formats quick experiments are easy to do with Claude Code. For example, I've always found it difficult to write a Y-statement as a shorter alternative to the typical decision records in markdown.

1

Hans L’Hoest @hanlho.com · Sep 2

Been experimenting with having Claude Code help me write decision records in ADR format after our technical discussions and use them as basis for further implementation.

hans.lhoest.eu/grounded-dec...

Grounded decision records from AI conversations

If you've read some of my posts before or worked with me, you know I like using Architectural Decision Records (ADRs) for lots of reasons. To me the most important one is documenting the why of a decision. If you've worked with AI models before, you'...

hans.lhoest.eu

Hans L’Hoest @hanlho.com · Aug 25

ccstatusline: very helpful to get insight in Claude Code's context statistics. Most useful for now: how much time is left in the 5h window (Block Time)?

github.com/sirmalloc/cc...

1

Hans L’Hoest @hanlho.com · Aug 22

Wrote an article:
hans.lhoest.eu/how-to-use-j...

#jj-vcs #claude-code

How to Use Jujutsu (jj) with Claude Code

I prefer Jujutsu (jj) over Git for version control, but getting Claude Code to work with jj proved tricky since Claude Code defaults to git. This post explains how to configure Claude Code to use jj instead of Git for version control operations and f...

hans.lhoest.eu

Hans L’Hoest @hanlho.com · Aug 22

...

```
2. If output is "jj" → use jj commands
3. If output is "git" → use git commands
4. If output is "none" → not a repository, inform user
```

Not sure if this actually a good idea but works so far.
(Note: system prompt contains more related `jj` instructions)

Hans L’Hoest @hanlho.com · Aug 22

TIL: you can add 'logic' directly to Claude Code's system prompt.

E.g. to detect which VCS:

```
Before ANY commit-related commands, you MUST:
1. Run `bash: if jj root >/dev/null 2>&1; then echo "jj"; elif git rev-parse --git-dir >/dev/null 2>&1; then echo "git"; else echo "none"; fi`
```

...

1

Hans L’Hoest @hanlho.com · Aug 20

Software architecture humblebundle

www.humblebundle.com/books/softwa...

Humble Tech Book Bundle: Software Architecture 2025 by O'Reilly

Pay what you want for the latest Humble Tech Bundle: Software Architecture 2025 and help support Code for America with your purchase!

www.humblebundle.com

Hans L’Hoest @hanlho.com · Aug 20

Add to CLAUDE .md:

```
**CRITICAL: This repository uses Jujutsu (jj), not git.** Always use `jj` commands for version control operations.

- Use `jj st --no-pager` to check status
- Use `jj log --no-pager` to view history
- Use `jj commit -m "message"` to create commits
```

#jj-vcs

Hans L’Hoest @hanlho.com · Aug 20

Claude Code has git pretty much baked into its system prompt.

Tried quite a few things to make Jujutsu ('jj') work with Claude Code and this seems to work:

1