Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Install Loop Engineering into a project via the unified CLI front door (@cobusgreyling/loop). Prefer this over invoking loop-init / loop-audit separately. Week-one is report-only; never enable auto-merge or unattended fixes unless the human explicitly asks and doctor is healthy.
.claude/skills/cobusgreyling-install-loop/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-07 | ✗→✓ | ▲ Improved | -33% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -46% | 0% |
| case-09 | ✗→✓ | ▲ Improved | -53% | 0% |
| case-10 | ✗→✓ | ▲ Improved | -43% | 0% |
| case-11 | ✗→✓ | ▲ Improved | -37% | 0% |
Leave the repo with a scaffolded report-only loop, a Loop Ready score, and clear next actions — using the single CLI:
bashnpx @cobusgreyling/loop
Do not deprecate or remove existing loop-init / loop-audit usages you find; both doors stay valid. Prefer loop for new work.
LOOP.md or STATE.md already exists, run doctor only (skip init unless human asks to re-scaffold):bash npx @cobusgreyling/loop doctor . --json npx @cobusgreyling/loop status .
daily-triage)| Pain | Pattern | |------|---------| | Morning chaos / unclear priorities | daily-triage | | PRs stalling | pr-babysitter | | CI red / flakes | ci-sweeper | | CVE / Dependabot noise | dependency-sweeper | | Post-merge TODOs | post-merge-cleanup | | Stale release notes | changelog-drafter | | Noisy issues | issue-triage | | One GitHub Action, almost no files | thin-loop |
claude | grok | codex | opencode (default claude).bash npx @cobusgreyling/loop init . --pattern <pattern> --tool <tool> Optional harness (only if human wants Foundry): add --with-foundry.
bash npx @cobusgreyling/loop doctor .
0 = healthy · 1 = warnings · 2 = blocked bash npx @cobusgreyling/loop cost -p <pattern> -l L1 -c 1d
/loop or scheduler command from init output.npx @cobusgreyling/loop badge .These remain fully supported (do not rewrite existing docs/scripts unless asked):
bashnpx @cobusgreyling/loop-init . npx @cobusgreyling/loop-audit . --suggest
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-20 | pass→pass | 9,100 | 10,210 | +12% | 1 | 1 | 0% | 1,489 | 1,548 | +4% | 0 | 0 | — |
case-01 | fail→fail | 11,526 | 16,844 | +46% | 1 | 1 | 0% | 502 | 1,176 | +134% | 0 | 0 | — |
case-02 | fail→fail | 10,596 | 15,431 | +46% | 1 | 1 | 0% | 220 | 945 | +330% | 0 | 0 | — |
case-03 | fail→fail | 4,576 | 16,604 | +263% | 1 | 1 | 0% | 180 | 1,069 | +494% | 0 | 0 | — |
case-04 | pass→fail | 11,150 | 28,101 | +152% | 1 | 1 | 0% | 1,271 | 3,468 | +173% | 0 | 0 | — |
case-05 | pass→fail | 13,338 | 13,764 | +3% | 1 | 1 | 0% | 1,329 | 2,230 | +68% | 0 | 0 | — |
case-06 | pass→fail | 9,575 | 17,231 | +80% | 1 | 1 | 0% | 1,875 | 2,943 | +57% | 0 | 0 | — |
case-07 | fail→pass | 11,084 | 8,361 | -25% | 1 | 1 | 0% | 1,852 | 1,238 | -33% | 0 | 0 | — |
case-08 | fail→pass | 14,629 | 8,443 | -42% | 1 | 1 | 0% | 2,378 | 1,296 | -46% | 0 | 0 | — |
case-09 | fail→pass | 15,907 | 4,410 | -72% | 1 | 1 | 0% | 2,757 | 1,286 | -53% | 0 | 0 | — |
case-10 | fail→pass | 13,340 | 3,411 | -74% | 1 | 1 | 0% | 2,301 | 1,309 | -43% | 0 | 0 | — |
case-11 | fail→pass | 17,362 | 8,263 | -52% | 1 | 1 | 0% | 2,034 | 1,280 | -37% | 0 | 0 | — |
case-12 | fail→pass | 11,405 | 4,677 | -59% | 1 | 1 | 0% | 2,047 | 1,478 | -28% | 0 | 0 | — |
case-13 | fail→pass | 9,065 | 7,522 | -17% | 1 | 1 | 0% | 1,430 | 1,210 | -15% | 0 | 0 | — |
case-14 | fail→pass | 8,961 | 2,744 | -69% | 1 | 1 | 0% | 1,773 | 1,151 | -35% | 0 | 0 | — |
case-15 | fail→pass | 9,502 | 3,645 | -62% | 1 | 1 | 0% | 1,599 | 1,230 | -23% | 0 | 0 | — |
case-16 | fail→pass | 17,105 | 11,580 | -32% | 1 | 1 | 0% | 3,237 | 1,808 | -44% | 0 | 0 | — |
case-17 | pass→pass | 30,718 | 7,034 | -77% | 1 | 1 | 0% | 5,156 | 1,046 | -80% | 0 | 0 | — |
case-18 | fail→pass | 10,304 | 7,313 | -29% | 1 | 1 | 0% | 1,684 | 1,100 | -35% | 0 | 0 | — |
case-19 | fail→pass | 15,829 | 2,481 | -84% | 1 | 1 | 0% | 1,802 | 1,142 | -37% | 0 | 0 | — |
case-21 | fail→pass | 20,979 | 9,489 | -55% | 1 | 1 | 0% | 2,212 | 1,400 | -37% | 0 | 0 | — |
case-22 | fail→pass | 17,689 | 8,483 | -52% | 1 | 1 | 0% | 2,000 | 1,254 | -37% | 0 | 0 | — |
case-23 | fail→pass | 27,342 | 10,067 | -63% | 1 | 1 | 0% | 1,197 | 1,302 | +9% | 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. 23 cases were attempted, and 20 counted toward the lift figure. The other 3 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 +52 percentage points is the difference between those two pass rates over the 20 comparable cases. 3 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.
| Model | Method | Date | Lift |
|---|---|---|---|
| gemini-3.6-flash | verified | 8/11/2026 | +64% |
Other measured skills in the registry, with their headline benchmark lift.