Github with the Halloween themed contribution graph. Would be nice to keep this version permanently :)
21
2nd time @clerk has a major outage in the last couple months. Maybe we should be self-hosting auth after all.
running my first long ( > 1 hour) codex /goal.
it's now 13+ hours in and still seemingly working towards the correct goal. Hasn't gone astray yet. We'll see if the result meets at least 70% of my expectation.
55
Woke up to one of my Anthropic API keys actively being used at about $30 every 10 minutes. Turns out it was from the @braintrust hack 😭 . Stopped it about an hour in (when I woke up), $189 in damages. Any recourse, or just par for the course?
First time this year that Opus is not my primary coding model. GPT-5.5 is now my preference for 70% of tasks.
Still using Opus every day (4.6 though), but if i'm working on something complex, it's GPT 5.5 all day.
64
I love great design but I'm not skilled at it. Claude design, Imagen 2.0, and design.md are incredible enablers.
As a developer (using AI agents for coding), I'd argue these design tools will have as big an impact on my work.
2
116
Claude Design is really good. But one-shot designs quickly converge to the same Claude AI aesthetic. If you stop there, you'll look like every other site in a few weeks. Push past the first output. The upside is exploring many directions.
Iteration cycle = killer feature
1
77
🚨 CRITICAL: Active supply chain attack on axios -- one of npm's most depended-on packages.
The latest [email protected] now pulls in [email protected], a package that did not exist before today. This is a live compromise.
This is textbook supply chain installer malware. axios has 100M+ weekly downloads. Every npm install pulling the latest version is potentially compromised right now.
Socket AI analysis confirms this is malware. plain-crypto-js is an obfuscated dropper/loader that:
• Deobfuscates embedded payloads and operational strings at runtime
• Dynamically loads fs, os, and execSync to evade static analysis
• Executes decoded shell commands
• Stages and copies payload files into OS temp and Windows ProgramData directories
• Deletes and renames artifacts post-execution to destroy forensic evidence
If you use axios, pin your version immediately and audit your lockfiles. Do not upgrade.
Anthropic should let Max subs work with any agent harness. The harness is becoming deeply workflow-specific, and Claude CLI just can't cover every style.
I get the concern with OpenClaw, but for developer agent harnesses, the API is way too expensive vs. a Max subscription.
2
5
335
Claude Code users can turn on "Fast" mode via /fast (first you have to allow /extra_usage). You get $50 in free credits.
/fast credits go FAST! I burned through $50 in less than an hour. But damn, it really enables better flow state.
2
107
Antigravity and Gemini 3 pro seem legit. As a Claude Code user, this is the first tool that feels competitive. Still early, still testing.
Current Downside: frequent error:
Agent execution terminated due to model provider overload. Please try again later.
PSA for Claude Code users:
Remember how in 1.x we'd constantly prompt "think deeply" or "think ultra hard"?
Well, in the new version, you just hit TAB to enter thinking mode!
Spent a whole day wondering why those nudges weren't as effective anymore🤦♂️
1
1
178
Love the new claude code /usage command. It's good to know when you might be close to hitting the limit, even on max.
Claude sonnet 4.5 starts new sessions using 56K of available context (without MCPs) before your first request. Seems like the majority (45K) is reserved to successfully autocompact, which has been a bit buggy of late.
I was already struggling w/ 200K, now i have less 😭
85
The bottleneck with ai-assisted coding is now overwhelmingly the ability to make good architectural and design decisions and to review code well (for longterm project stability).
Frankly, those are skills a lot of developers lack.
Zed's vision for their DeltaDB approach is really intriguing.
Github commits do at times feel "too macro" for the current fast-paced, fast-experimental, and agent collaborative workflows we're all in.
zed.dev/blog/sequoia-backs-z…
1
75
I really like how this brand presents its ingredients list (protein bar). US brands should adopt this, though for many products they would need to use really small fonts.
1
50






