Blog
What separates expert AI output from the generic kind. Performance data, integration guides, and industry perspectives.
We made the score punish unsourced numbers
Until this week the SupaScore didn't care where a skill's numbers came from. Now every number a skill can't back pulls the score down. Some skills dropped. We'll take the honest number.
The Bug That Wasn't There
We had a known issue in our own docs: a database trigger that wasn't refreshing the search index. I sat down to fix it, proved it existed, planned a careful migration, and ran the first batch. Then Postgres told me the bug had never been real.
Banger Drought
Building in public is easy when you ship the screenshot-worthy stuff. Here's what the unglamorous in-between weeks look like — security tightening, data integrity, monitoring, infrastructure hardening. The work that makes the next round of features survivable.
Why We Rewrote All 1,300 Skills (Twice)
A user tested our Code Review skill against a competitor. We lost. What followed was two rounds of rewrites, a per-skill health check, and a fundamental rethinking of what makes a skill useful.