Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Read and write Obsidian vault files, manage wiki-links, process markdown with YAML frontmatter. Use when working with vault file operations, creating notes, or managing links.
.claude/skills/ballred-obsidian-vault-ops/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | -2% | 0% |
| case-03 | ✗→✓ | ▲ Improved | -33% | 0% |
| case-04 | ✗→✓ | ▲ Improved | -31% | 0% |
| case-10 | ✗→✓ | ▲ Improved | -31% | 0% |
| case-11 | ✗→✓ | ▲ Improved | -36% | 0% |
Core operations for reading, writing, and managing files in an Obsidian vault.
vault-root/
├── CLAUDE.md # Main context (always read first)
├── Daily Notes/ # YYYY-MM-DD.md format
├── Goals/ # Goal cascade files
├── Projects/ # Project folders with CLAUDE.md
├── Templates/ # Reusable note structures
└── Archives/ # Completed/inactive content*.md, Daily Notes/*.mdmarkdown[[Note Name]] # Simple link [[Note Name|Display Text]] # Link with alias [[Note Name#Section]] # Link to section
Standard frontmatter structure:
yaml--- date: 2024-01-15 tags: [tag1, tag2] status: active ---
When processing templates, replace:
{{date}} - Today's date (YYYY-MM-DD){{date:format}} - Formatted date{{date-1}} - Yesterday{{date+1}} - Tomorrow{{time}} - Current timeDaily Notes/{date}.md existsTemplates/Daily Template.mdDaily Notes/{date}.md#priority/high, #priority/medium, #priority/low#active, #waiting, #completed, #archived#work, #personal, #health, #learning| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | pass→pass | 7,475 | 4,873 | -35% | 1 | 1 | 0% | 1,332 | 1,581 | +19% | 0 | 0 | — |
case-02 | fail→pass | 10,816 | 7,488 | -31% | 1 | 1 | 0% | 2,013 | 1,976 | -2% | 0 | 0 | — |
case-03 | fail→pass | 9,940 | 3,419 | -66% | 1 | 1 | 0% | 1,932 | 1,294 | -33% | 0 | 0 | — |
case-04 | fail→pass | 9,036 | 2,965 | -67% | 1 | 1 | 0% | 1,738 | 1,206 | -31% | 0 | 0 | — |
case-05 | pass→pass | 3,588 | 2,541 | -29% | 1 | 1 | 0% | 663 | 1,034 | +56% | 0 | 0 | — |
case-06 | pass→pass | 3,649 | 1,637 | -55% | 1 | 1 | 0% | 727 | 826 | +14% | 0 | 0 | — |
case-07 | pass→pass | 3,507 | 2,460 | -30% | 1 | 1 | 0% | 717 | 1,141 | +59% | 0 | 0 | — |
case-08 | pass→pass | 5,093 | 2,520 | -51% | 1 | 1 | 0% | 889 | 1,076 | +21% | 0 | 0 | — |
case-09 | pass→pass | 8,351 | 3,182 | -62% | 1 | 1 | 0% | 1,544 | 1,206 | -22% | 0 | 0 | — |
case-10 | fail→pass | 8,089 | 2,500 | -69% | 1 | 1 | 0% | 1,484 | 1,030 | -31% | 0 | 0 | — |
case-11 | fail→pass | 8,933 | 2,446 | -73% | 1 | 1 | 0% | 1,538 | 987 | -36% | 0 | 0 | — |
case-12 | fail→pass | 11,361 | 1,869 | -84% | 1 | 1 | 0% | 1,997 | 873 | -56% | 0 | 0 | — |
case-13 | pass→pass | 5,235 | 1,868 | -64% | 1 | 1 | 0% | 927 | 875 | -6% | 0 | 0 | — |
case-14 | fail→pass | 12,562 | 5,198 | -59% | 1 | 1 | 0% | 2,105 | 1,578 | -25% | 0 | 0 | — |
case-15 | fail→fail | 7,017 | 6,036 | -14% | 1 | 1 | 0% | 1,296 | 1,743 | +34% | 0 | 0 | — |
case-16 | fail→fail | 11,372 | 5,963 | -48% | 1 | 1 | 0% | 1,896 | 1,596 | -16% | 0 | 0 | — |
case-17 | pass→pass | 6,699 | 2,513 | -62% | 1 | 1 | 0% | 1,203 | 1,031 | -14% | 0 | 0 | — |
case-18 | pass→pass | 11,016 | 1,995 | -82% | 1 | 1 | 0% | 1,756 | 929 | -47% | 0 | 0 | — |
case-19 | pass→pass | 5,461 | 1,493 | -73% | 1 | 1 | 0% | 958 | 848 | -11% | 0 | 0 | — |
case-20 | pass→pass | 9,216 | 9,193 | -0% | 1 | 1 | 0% | 1,842 | 2,776 | +51% | 0 | 0 | — |
case-21 | pass→pass | 9,168 | 9,067 | -1% | 1 | 1 | 0% | 1,882 | 2,403 | +28% | 0 | 0 | — |
case-22 | pass→pass | 14,318 | 13,958 | -3% | 1 | 1 | 0% | 2,585 | 3,336 | +29% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted. The headline lift of +32 percentage points is the difference between those two pass rates over the 22 comparable cases.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.