CTO. Long-time PHP Core dev & OSS contributor. I make things faster, more secure and scalable, now with AI in the loop. My projects @ github.com/iliaal
Hey @cursor_ai why am I getting this when analyzing my own log files?
Error: Request blocked
We are unable to complete this request because it triggered restrictions on cyber content and was blocked under the model provider's usage guidelines. Try a less sensitive prompt.
whetstone 4.6.0 adds an optional judge to the subagent skill-injection hook. The obvious version of that feature lets a model fill the empty slots. That is how you end up with five skills injected into every subagent call.
So the hook asks two questions. Keyword tiers pick up to
Got access to @typesafeai first experiment was testing whether Jev can choose the right Whetstone skill for a request.
- On newly written, tricky requests: Jev got 63 of 64 right. Which is much better than 32/64 regex rules got
- On older test cases: regex rules got all 132
Somethings claude (Opus 5) says the strangest things
"This one came out of Codex's dying breath and is a live crash in my own merged fix"
I am using codex plugin for claude as reviewer for Opus' work
