Experiment Engine logo

Experiment Engine

Community
vibeeval
experiment-engine

Otonom deney dongusu. Kod degisikligi yap, olc, karsilastir, kabul et veya geri al. Metrik bazli karar verme ile performans, boyut veya kalite optimizasyonu. Tek basina veya agent ile kullan.

Overview

Publishervibeeval
Repositoryvibecosystem
Skill nameexperiment-engine
Stars
533
Forks
44
Bundled files
Instructions only
LicenseMIT
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • Self-contained

    Everything the model needs lives in the instructions — no extra files to sync.

  • Open source

    Published by vibeeval on GitHub. Read the source before you install it.

Installation

Install the Experiment Engine AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
    https://github.com/vibeeval/vibecosystem/tree/main/skills/experiment-engine
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/vibeeval/vibecosystem.git /tmp/vibecosystem
mkdir -p .claude/skills
cp -r /tmp/vibecosystem/skills/experiment-engine .claude/skills/experiment-engine
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Experiment Engine in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Experiment Engine on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Experiment Engine is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

Not / Overlap: Bu skill experiment-loop ile ayni otonom deney dongusu kavramini kapsar (bu dosya TR, experiment-loop EN). Yeni isler icin experiment-loop tercih edilir; v4.0 da birlestirilecek.

Experiment Engine

Bir hedef belirle, sistematik olarak deneyler yap, sadece iyilestirenleri tut.

Core Loop

HEDEF BELIRLE
  └─ "API response time'i %20 dusur"
  └─ "Bundle size'i 500KB'nin altina getir"
  └─ "Test coverage'i %90'a cikar"

BASELINE OLC
  └─ Mevcut metrigi kaydet (ornek: 340ms, 720KB, %78)

DENEY DONGUSU (N kez tekrarla):
  ┌─────────────────────────────────────┐
  │ 1. MODIFY  - Tek degisiklik yap    │
  │ 2. VERIFY  - Metrigi olc           │
  │ 3. COMPARE - Baseline ile kiyasla  │
  │ 4. DECIDE  - Kabul / Reddet        │
  │    ├─ Iyilesti → COMMIT + yeni     │
  │    │             baseline           │
  │    └─ Kotulesti → ROLLBACK         │
  └─────────────────────────────────────┘

RAPOR OLUSTUR
  └─ N deney, X kabul, Y red, final metrik

Kullanim Alanlari

Performans Optimizasyonu

bash
# Hedef: API response time < 200ms
# Baseline: 340ms

# Deney 1: Database query'ye index ekle
git stash  # mevcut durumu kaydet
# ... index ekle ...
curl -w "%{time_total}" http://localhost:3000/api/users  # 280ms
# 340ms -> 280ms = IYILESTI → COMMIT

# Deney 2: Response'u cache'le
# ... Redis cache ekle ...
curl -w "%{time_total}" http://localhost:3000/api/users  # 45ms
# 280ms -> 45ms = IYILESTI → COMMIT

# Deney 3: JSON serializer degistir
# ... fast-json-stringify ekle ...
curl -w "%{time_total}" http://localhost:3000/api/users  # 42ms
# 45ms -> 42ms = MINIMAL IYILESME → REDDET (karmasiklik artmaya degmez)

# Sonuc: 340ms -> 45ms (%87 iyilesme), 2/3 deney kabul edildi

Bundle Size Azaltma

bash
# Hedef: < 500KB
# Baseline olc
BASELINE=$(npx next build 2>&1 | grep "First Load JS" | awk '{print $4}')

# Deney dongusu
experiments=(
  "lodash yerine lodash-es"
  "moment yerine dayjs"
  "tree-shaking acik mi kontrol"
  "dynamic import lazy component'ler"
  "image optimize (next/image)"
)

for exp in "${experiments[@]}"; do
  echo "=== Deney: $exp ==="
  # degisiklik yap...
  NEW=$(npx next build 2>&1 | grep "First Load JS" | awk '{print $4}')
  if [ "$NEW" -lt "$BASELINE" ]; then
    echo "KABUL: $BASELINE -> $NEW"
    BASELINE=$NEW
    git add -A && git stash  # kaydet
  else
    echo "RED: $NEW >= $BASELINE"
    git checkout .  # geri al
  fi
done

Test Coverage Artirma

bash
# Hedef: %90 coverage
# Baseline
BASELINE=$(npx jest --coverage --silent 2>&1 | grep "All files" | awk '{print $4}')

# Her dosya icin test yaz, coverage'i olc
for file in $(find src -name "*.ts" -not -name "*.test.*"); do
  echo "=== Test: $file ==="
  # test yaz...
  NEW=$(npx jest --coverage --silent 2>&1 | grep "All files" | awk '{print $4}')
  if (( $(echo "$NEW > $BASELINE" | bc -l) )); then
    echo "KABUL: %$BASELINE -> %$NEW"
    BASELINE=$NEW
  fi
done

Deney Protokolu

Tek Degisiklik Kurali

YANLIS: Ayni anda 3 sey degistirip "daha hizli oldu" demek
  → Hangi degisiklik etkili oldugunu bilemezsin

DOGRU: Her seferinde TEK degisiklik yap
  → Neyin ise yaradigini kesin bilirsin

Rollback Stratejisi

bash
# Yontem 1: git stash (basit)
git stash         # deney oncesi
# ... deney ...
git stash pop     # basarisizsa geri al

# Yontem 2: git worktree (izole)
git worktree add /tmp/experiment-1 -b exp/perf-test
cd /tmp/experiment-1
# ... deney ...
# basarisizsa worktree'yi sil

# Yontem 3: checkpoint (karmasik deneyler)
git add -A && git commit -m "checkpoint: pre-experiment"
# ... deney ...
# basarisizsa: git reset --hard HEAD~1

Metrik Toplama

typescript
interface ExperimentResult {
  id: string
  description: string
  baseline: number
  result: number
  improvement: number  // yuzde
  accepted: boolean
  duration: number     // saniye
  timestamp: string
}

// Deney raporu
interface ExperimentReport {
  goal: string
  metric: string
  baselineValue: number
  finalValue: number
  totalExperiments: number
  accepted: number
  rejected: number
  totalImprovement: number  // yuzde
  experiments: ExperimentResult[]
}

Deney Sablonu

markdown
# Deney Raporu: [Hedef]

## Ozet
- Hedef: [metrik] < [esik]
- Baseline: [baslangic degeri]
- Final: [son deger]
- Iyilesme: [yuzde]
- Deneyler: [kabul]/[toplam]

## Deneyler

| # | Aciklama | Onceki | Sonraki | Degisim | Karar |
|---|----------|-------:|--------:|--------:|-------|
| 1 | Index ekle | 340ms | 280ms | -18% | KABUL |
| 2 | Redis cache | 280ms | 45ms | -84% | KABUL |
| 3 | JSON serializer | 45ms | 42ms | -7% | RED |

## Ogrenim
- En etkili: Redis cache (-84%)
- Degmez: JSON serializer degisimi (karmasiklik > kazanim)

Otomatik Deney Modu

bash
# experiment-loop.sh
# Kullanim: ./experiment-loop.sh "response_time" "200" "ms" 10

METRIC=$1       # olculecek metrik
TARGET=$2       # hedef deger
UNIT=$3         # birim
MAX_ROUNDS=$4   # max deney sayisi

ROUND=0
BASELINE=$(measure_$METRIC)

while [ $ROUND -lt $MAX_ROUNDS ]; do
  ROUND=$((ROUND + 1))

  # Claude'a optimize ettir
  claude -p "Optimize $METRIC. Current: ${BASELINE}${UNIT}. Target: <${TARGET}${UNIT}. Make ONE small change." --no-input

  # Olc
  NEW=$(measure_$METRIC)

  if [ "$NEW" -lt "$BASELINE" ]; then
    echo "Round $ROUND: KABUL ($BASELINE -> $NEW)"
    BASELINE=$NEW
    git add -A && git commit -m "experiment: $METRIC improved to ${NEW}${UNIT}"
  else
    echo "Round $ROUND: RED ($NEW >= $BASELINE)"
    git checkout .
  fi

  # Hedefe ulastik mi?
  if [ "$BASELINE" -le "$TARGET" ]; then
    echo "HEDEF ULASILDI: ${BASELINE}${UNIT} <= ${TARGET}${UNIT}"
    break
  fi
done

vibecosystem Entegrasyonu

  • profiler agent: Performans deneylerinde metrik toplama
  • nitro agent: Optimization deneylerini yonetme
  • tdd-guide agent: Coverage deneylerinde test yazma
  • verifier agent: Her deney sonrasi build/test dogrulama
  • self-learner agent: Basarili deneyleri pattern olarak kaydet
  • experiment-loop skill: Bu skill'in mevcut complementary'si

Frequently asked questions

What does the Experiment Engine AI skill do?

Otonom deney dongusu. Kod degisikligi yap, olc, karsilastir, kabul et veya geri al. Metrik bazli karar verme ile performans, boyut veya kalite optimizasyonu. Tek basina veya agent ile kullan.

Why use Experiment Engine on TypingMind?

Because you install it once and use it with any model. Experiment Engine is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Experiment Engine in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/vibeeval/vibecosystem/tree/main/skills/experiment-engine. TypingMind reads its SKILL.md and installs it as a skill you can enable per chat.

Which AI models can use Experiment Engine?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Experiment Engine?

As many as you like. As long as a model supports skills, you can use Experiment Engine with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Experiment Engine AI skill free?

Yes. It is published on GitHub by vibeeval under the MIT license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇