Skills load for judgement calls too, even with many other skills installed.
- 50 skills (patch bump each): the "or asks" part of each description now covers the usual judgement calls in its area: choosing between two options ("which promise to lead with", "screen recording or AI avatar"), whether something is okay or fair ("is it fair to say a rival lacks offline mode", "is this markup allowed"), and whether it's enough ("are six directory listings enough"). Lower-value list items were cut to pay for it, so the pack stays at 18,662 characters and all 53 skills keep their descriptions with 90 other skills installed. On judgement-style prompts nobody read while writing the change, the right skill was picked for about 90% (92.5% twice, 89.6% for the shipped wording, from 66.5%) with 90 other skills installed and 95.3% (from 85.8%) with the pack alone, with no false triggers. programmatic-seo, revops, and signup are unchanged.
- Trigger eval: a fourth held-out set (
heldout4.json, judgement-style), used unread to judge this change.
Skill ZIP assets
Complete source catalog: 53 skills.
Claude web target: 53 skills with reviewed short descriptions.
Perplexity Computer target: 53 skills.
Choose <name>-claude-web.zip for the documented single-folder web upload format.
Choose <name>-perplexity.zip for Perplexity Computer (SKILL.md at the archive root).
Source ZIPs retain original metadata; they are not substitutes for omitted web ZIPs.
Live client upload, account limits, and invocation behavior have not been verified.
Web omissions (also recorded in claude-web-manifest.json):
- None.