3 Commits

Author	SHA1	Message	Date
Claude	8fc3803385	Improve skill accuracy, completeness, and eval coverage ... - Fix README command count (100+ → 130+) to match SKILL.md - Add 9 missing command groups to README table (bookmarks, themes, snippets, commands/hotkeys, bases, history, workspace, diff, and expanded dev/vault groups with all subcommands) - Expand README troubleshooting table with 4 entries matching SKILL.md (multi-vault naming, property:set list type, Windows .com issues, Git Bash / MSYS2 colon-param exit 127) - Enhance SKILL.md frontmatter: add version field, expand description with explicit skip conditions (iCloud sync, Dataview syntax, keyboard shortcuts, external scripts), and add 22-entry triggers list for more reliable skill activation across platforms - Expand eval set from 20 → 35 cases: 10 new positive cases covering sync restore, bookmarks, eval, backlinks, screenshots, tag counts, Bases queries, recents, permanent delete, and rename; 5 new negative cases covering graph view explanation, publish options, hotkey config, unresolved links concept, and Templater plugin syntax https://claude.ai/code/session_01285FCGJmPvrLodTg4nVuc9	2026-03-05 20:52:00 +00:00
pablo-mano	a882e74591	Apply optimized description from run_loop eval; add v2 eval set ... Description rewritten to emphasize vault action intent over CLI syntax, with explicit "without this skill Claude has no vault access" framing. Adds v2 eval set (20 queries built around natural-language vault requests). Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>	2026-03-04 00:37:19 +01:00
pablo-mano	c7b35b314d	Optimize skill description and add eval set ... Replace the broad description with a sharper, action-oriented one that emphasizes the terminal/script context as the trigger signal and explicitly lists concrete subcommands. Add eval/ directory with 20 trigger eval queries and the HTML reviewer used to generate them. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>	2026-03-04 00:10:04 +01:00