Strong Opinions
Perspectives and takes on the developer landscape.
Category
Strong Opinions
Tags
46 more tagsFewer tags
Newsletter
New posts land in your inbox when they publish. No spam, unsubscribe anytime.
You're in. New posts head your way.
Filters1
Category
Strong Opinions
Tags
46 more tagsFewer tags
64 posts

Grok 4.6 tied Sol and still lost the terminal
Grok 4.6 Terminal-Bench is the row that should move the buy. It tied Sol at 61 on the index and still sat at 26 percent on v3.0. Shop a named agent loop.

Skill scanning is antivirus for markdown
Skill scanning checks a third-party zip on upload before it can run. Turn the Enterprise toggle on, then still open the folder and read allowed-tools.

Claude Code design mockups land before the code
Claude Code design mockups land before the code. Keep Main.dc.html in git so the order survives the session. The hosted canvas is a scrape you may not share.

Qwen 3.8 overthinks on the default reasoning setting
Qwen 3.8 overthinks on the default xhigh setting. A pelican SVG ate 22,276 thinking tokens and 21 minutes. Start on low or off, then raise it for hard jobs.

750 tokens a second does not make review faster
GPT-5.6 Ultrafast is 750 tok/s of author speed. That fills the human review queue harder. Pin the changelog to the bottleneck, then budget the review lane.

Auto mode is the default because you already approved everything
Claude Code auto mode is default for new Pro, Max, and Team sessions because you already approved 97% of prompts. Put deny and ask rules in front now.

Claude Code sessions can now message each other
Claude Code sessions can now message each other with a text ping. Use ListAgents for status. Keep the worktree. A ping is not a lock on a shared file.

Enterprise can now pull Claude Code transcripts off the laptop
Enterprise can now pull Claude Code transcripts off a company laptop. The Compliance API rebuilds them from API calls. Treat local as resume, not privacy.

The Hugging Face break-in was a swarm with a scratchpad
The Hugging Face break-in was a swarm with a scratchpad. Watch the shared eval cache, because a wipe only kills the process sitting in front of you now.