Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Generate AGENTS.md and AI configuration files for your project. Use when the user wants to create agent instructions, set up AI configs, or says "create AGENTS.md", "configure my AI assistant", or "generate agent files".
.claude/skills/khazp-vibe-agents/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | 345% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -72% | 0% |
| case-06 | ✗→✓ | ▲ Improved | -50% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -58% | 0% |
| case-10 | ✗→✓ | ▲ Improved | -59% | 0% |
Read the agreed PRD and Tech Design using vibe.project.json paths when available; otherwise inspect docs/PRD--MVP.md and docs/TechDesign--MVP.md. Reconcile conflicts and reuse Handoff Context before asking anything.
Run npx vibeworkflow --dry-run --json to inspect setup changes, then initialize within existing authorization. Existing files are preserved by default. Use --force only for explicitly intended replacement after showing affected files. The CLI installs selected files directly; it does not create a templates/ directory.
Fill required placeholders from decisions. AGENTS.md contains stable rules; MEMORY.md contains current progress; agent_docs contains concise build context. Run npx vibeworkflow doctor and resolve required setup errors. Build and behavior remain Not checked.
For a chat without filesystem access, use the repository's downloadable docs/context-pack.md plus the user's product documents; output clearly separated files to save. Do not invent absent template content. Tool configurations follow the installed client's supported settings; do not enable broad permissions automatically.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 13,630 | 16,702 | +23% | 1 | 1 | 0% | 256 | 637 | +149% | 0 | 0 | — |
case-02 | fail→pass | 8,875 | 19,150 | +116% | 1 | 1 | 0% | 662 | 2,946 | +345% | 0 | 0 | — |
case-03 | fail→fail | 7,613 | 9,111 | +20% | 1 | 1 | 0% | 265 | 476 | +80% | 0 | 0 | — |
case-04 | pass→pass | 11,698 | 7,710 | -34% | 1 | 1 | 0% | 1,231 | 726 | -41% | 0 | 0 | — |
case-05 | fail→pass | 14,286 | 6,790 | -52% | 1 | 1 | 0% | 1,959 | 552 | -72% | 0 | 0 | — |
case-06 | fail→pass | 20,933 | 7,846 | -63% | 1 | 1 | 0% | 1,446 | 722 | -50% | 0 | 0 | — |
case-07 | fail→pass | 12,510 | 6,872 | -45% | 1 | 1 | 0% | 1,321 | 560 | -58% | 0 | 0 | — |
case-08 | pass→pass | 14,798 | 6,312 | -57% | 1 | 1 | 0% | 1,695 | 457 | -73% | 0 | 0 | — |
case-09 | pass→pass | 15,444 | 7,102 | -54% | 1 | 1 | 0% | 1,800 | 559 | -69% | 0 | 0 | — |
case-10 | fail→pass | 14,266 | 7,017 | -51% | 1 | 1 | 0% | 1,372 | 557 | -59% | 0 | 0 | — |
case-11 | fail→pass | 17,325 | 7,139 | -59% | 1 | 1 | 0% | 2,056 | 580 | -72% | 0 | 0 | — |
case-23 | pass→pass | 13,977 | 15,045 | +8% | 1 | 1 | 0% | 1,707 | 1,977 | +16% | 0 | 0 | — |
case-12 | fail→pass | 14,910 | 7,282 | -51% | 1 | 1 | 0% | 1,601 | 620 | -61% | 0 | 0 | — |
case-13 | pass→pass | 14,499 | 7,356 | -49% | 1 | 1 | 0% | 1,523 | 537 | -65% | 0 | 0 | — |
case-14 | fail→pass | 21,304 | 7,255 | -66% | 1 | 1 | 0% | 2,391 | 568 | -76% | 0 | 0 | — |
case-15 | fail→pass | 20,887 | 8,833 | -58% | 1 | 1 | 0% | 2,504 | 908 | -64% | 0 | 0 | — |
case-16 | fail→pass | 14,705 | 8,565 | -42% | 1 | 1 | 0% | 1,544 | 718 | -53% | 0 | 0 | — |
case-17 | fail→pass | 15,461 | 8,343 | -46% | 1 | 1 | 0% | 1,613 | 655 | -59% | 0 | 0 | — |
case-18 | fail→pass | 19,163 | 11,831 | -38% | 1 | 1 | 0% | 2,316 | 1,472 | -36% | 0 | 0 | — |
case-19 | pass→pass | 16,374 | 7,354 | -55% | 1 | 1 | 0% | 1,678 | 626 | -63% | 0 | 0 | — |
case-20 | pass→pass | 16,826 | 8,392 | -50% | 1 | 1 | 0% | 1,790 | 790 | -56% | 0 | 0 | — |
case-21 | pass→pass | 20,648 | 11,595 | -44% | 1 | 1 | 0% | 2,404 | 1,249 | -48% | 0 | 0 | — |
case-22 | pass→pass | 14,809 | 12,812 | -13% | 1 | 1 | 0% | 1,818 | 1,655 | -9% | 0 | 0 | — |
case-24 | pass→pass | 15,790 | 16,068 | +2% | 1 | 1 | 0% | 2,063 | 2,367 | +15% | 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. 24 cases were attempted, and 22 counted toward the lift figure. The other 2 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 +50 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.
| Model | Method | Date | Lift |
|---|---|---|---|
| gemini-3.6-flash | verified | 8/26/2026 | +45% |
| gemini-3.6-flash | verified | 8/12/2026 | +48% |
Other measured skills in the registry, with their headline benchmark lift.