Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Install DB Browser for SQLite (if not already installed) and open a .sqlite file in it. macOS only.
.claude/skills/sonichi-db-browser/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-04 | ✗→✓ | ▲ Improved | -2% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -47% | 0% |
| case-15 | ✗→✓ | ▲ Improved | -52% | 0% |
| case-16 | ✗→✓ | ▲ Improved | -29% | 0% |
| case-17 | ✗→✓ | ▲ Improved | -44% | 0% |
Install DB Browser for SQLite via Homebrew if it isn't already installed, then open a given .sqlite file in it. Macos only — uses brew --cask + open -a.
Usage: /db-browser <path-to.sqlite>
ARGUMENTS: $ARGUMENTS
$(bash scripts/sutando-config.sh workspace)/data/conversation.sqlite (the M0 helper — reads sutando.config.local.json, defaulting to <repo>/workspace/data/conversation.sqlite; $SUTANDO_WORKSPACE no longer honored as of v0.8 / #1440). Bail with an error if the file doesn't exist.mdfind "kMDItemKind == 'Application'" 2>/dev/null | grep -qi 'DB Browser for SQLite' (or [ -d /Applications/DB\ Browser\ for\ SQLite.app ]). If installed, skip step 3.brew install --cask db-browser-for-sqlite. Requires brew. If brew is missing, stop and ask the user to install brew first.osascript -e 'tell application "DB Browser for SQLite" to quit'; sleep 2. Skip if not running.open -a "DB Browser for SQLite" "<resolved-path>". Sleep ~2s so the window is up before reporting back.lsof -c "DB Browser" 2>/dev/null | grep "<resolved-path>"..sqlite-wal sidecar holds them. DB Browser reads both — no checkpoint needed. (Run sqlite3 <db> 'PRAGMA wal_checkpoint(TRUNCATE);' only if you want a single-file dump.)| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 35,504 | 27,315 | -23% | 1 | 1 | 0% | 901 | 1,364 | +51% | 0 | 0 | — |
case-02 | fail→fail | 7,544 | 9,521 | +26% | 1 | 1 | 0% | 348 | 952 | +174% | 0 | 0 | — |
case-03 | fail→fail | 10,649 | 10,915 | +2% | 1 | 1 | 0% | 1,753 | 1,056 | -40% | 0 | 0 | — |
case-04 | fail→pass | 9,076 | 5,231 | -42% | 1 | 1 | 0% | 1,497 | 1,473 | -2% | 0 | 0 | — |
case-05 | pass→pass | 11,229 | 5,249 | -53% | 1 | 1 | 0% | 1,987 | 1,515 | -24% | 0 | 0 | — |
case-06 | pass→pass | 2,809 | 1,651 | -41% | 1 | 1 | 0% | 460 | 835 | +82% | 0 | 0 | — |
case-07 | fail→pass | 13,334 | 3,467 | -74% | 1 | 1 | 0% | 2,140 | 1,143 | -47% | 0 | 0 | — |
case-08 | pass→pass | 11,384 | 2,436 | -79% | 1 | 1 | 0% | 682 | 991 | +45% | 0 | 0 | — |
case-09 | pass→pass | 4,682 | 2,045 | -56% | 1 | 1 | 0% | 849 | 996 | +17% | 0 | 0 | — |
case-10 | pass→pass | 12,244 | 4,201 | -66% | 1 | 1 | 0% | 2,031 | 1,238 | -39% | 0 | 0 | — |
case-11 | pass→pass | 10,456 | 3,414 | -67% | 1 | 1 | 0% | 1,684 | 1,214 | -28% | 0 | 0 | — |
case-12 | fail→fail | 5,193 | 13,527 | +160% | 1 | 1 | 0% | 893 | 1,828 | +105% | 0 | 0 | — |
case-13 | pass→fail | 6,170 | 6,837 | +11% | 1 | 1 | 0% | 947 | 1,063 | +12% | 0 | 0 | — |
case-14 | pass→fail | 5,570 | 8,020 | +44% | 1 | 1 | 0% | 1,024 | 1,136 | +11% | 0 | 0 | — |
case-15 | fail→pass | 11,510 | 1,710 | -85% | 1 | 1 | 0% | 1,784 | 860 | -52% | 0 | 0 | — |
case-16 | fail→pass | 10,772 | 2,641 | -75% | 1 | 1 | 0% | 1,492 | 1,054 | -29% | 0 | 0 | — |
case-17 | fail→pass | 15,515 | 4,477 | -71% | 1 | 1 | 0% | 2,268 | 1,270 | -44% | 0 | 0 | — |
case-18 | fail→pass | 9,195 | 2,753 | -70% | 1 | 1 | 0% | 1,341 | 1,102 | -18% | 0 | 0 | — |
case-19 | fail→pass | 11,514 | 1,671 | -85% | 1 | 1 | 0% | 1,756 | 830 | -53% | 0 | 0 | — |
case-20 | pass→pass | 4,070 | 7,120 | +75% | 1 | 1 | 0% | 587 | 1,757 | +199% | 0 | 0 | — |
case-21 | pass→pass | 6,579 | 4,437 | -33% | 1 | 1 | 0% | 1,106 | 1,381 | +25% | 0 | 0 | — |
case-22 | pass→pass | 4,878 | 5,759 | +18% | 1 | 1 | 0% | 783 | 1,577 | +101% | 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, and 16 counted toward the lift figure. The other 6 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +23 percentage points is the difference between those two pass rates over the 16 comparable cases. 5 cases got worse with the skill loaded, and they are included in that figure.
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.