メニュー
This skill should be used when validating features end-to-end without mocks, testing integrations, or when "scenario test", "e2e test", or "no mocks" are mentioned.
デフォルトでは、最初にソースを確認する Prompt が選択されています。直接コマンドに切り替えるか、ローカルコピーをダウンロードすることもできます。
インストールを決める前に、SKILL.md と SkillsMP に表示されている付属ファイルをお読みください。
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
直接コマンドでは確認用 Prompt が省略されます。実行前にソースを確認してください。
npx skills add https://github.com/outfitter-dev/agents --skill scenariosコマンドは1行のまま表示されます。コピー前に横へスクロールして全体を確認してください。
ローカルで確認しますか？SkillsMP が現在取得できるファイルをダウンロードできます。
This skill should be used when the user asks to "complete a branch", "merge to main", "finish my feature", "ship this branch", "integrate to main", "create a PR from GitButler", or when `--complete-branch` flag is mentioned. Guides completion of GitButler virtual branches with safety snapshots, integration workflows, and cleanup.
This skill should be used when coordinating multiple AI agents working concurrently, handling agent handoffs, transferring commits between agents, or when "multi-agent", "concurrent agents", "parallel agents", "agent collaboration", or "parallel execution" are mentioned with GitButler. Provides virtual branch patterns for parallel execution without coordination overhead.
This skill should be used when creating stacks, dependent branches, or when "stack", "stacked branches", "anchor", "--anchor", "but branch new -a", "create dependent branch", or "break feature into PRs" are mentioned with GitButler. Covers anchor-based stacking for dependent features and reviewable PR breakdown.
SKILL.md を表示中
| name | scenarios |
| description | This skill should be used when validating features end-to-end without mocks, testing integrations, or when "scenario test", "e2e test", or "no mocks" are mentioned. |
| metadata | {"version":"1.0.0"} |
End-to-end validation using real dependencies, no mocks ever.
<when_to_use>
NOT for: unit testing, mock testing, performance benchmarking, load testing
</when_to_use>
<iron_law>
NO MOCKS EVER.
Truth hierarchy:
Mocks test your assumptions, not reality. When mocks pass but production fails, the mock lied. When scenarios fail, reality spoke.
Test against real databases, real APIs, real services. Use test credentials, staging environments, local instances — but always real implementations.
</iron_law>
<directory_structure>
Throwaway test scripts for quick validation. Self-contained, runnable, disposable.
CRITICAL: Verify .scratch/ in .gitignore before first use.
Successful scenario patterns documented as JSONL. One scenario per line, each a complete JSON object.
Purpose: capture proven patterns, regression indicators, reusable test cases.
Structure:
{"name":"auth-login-success","description":"User logs in with valid credentials","setup":"Create test user with known password","steps":["POST /auth/login with credentials","Receive JWT token","GET /auth/me with token"],"expected":"User profile returned with correct data","tags":["auth","jwt","happy-path"]}
{"name":"auth-login-invalid","description":"Login fails with wrong password","setup":"Test user exists","steps":["POST /auth/login with wrong password"],"expected":"401 Unauthorized, no token issued","tags":["auth","error-handling"]}
</directory_structure>
<scratch_directory>
Quick validation without ceremony. Write script, run against real deps, verify behavior, delete or document.
test-{feature}.ts — feature validation (test-auth-flow.ts)debug-{issue}.ts — investigate specific bug (debug-token-expiry.ts)prove-{behavior}.ts — demonstrate expected behavior (prove-rate-limiting.ts)explore-{api}.ts — learn external API behavior (explore-stripe-webhooks.ts)// .scratch/test-auth-flow.ts
import { db } from '../src/db'
import { api } from '../src/api'
async function testAuthFlow() {
// Setup: real test user in real database
const user = await db.users.create({
email: 'test@example.com',
password: 'hashed-test-password'
})
// Execute: real HTTP requests
const loginRes = await api.post('/auth/login', {
email: user.email,
password: 'test-password'
})
// Verify: actual response
console.assert(loginRes.status === 200, 'Login should succeed')
console.assert(loginRes.body.token, 'Should receive JWT token')
const meRes = await api.get('/auth/me', {
headers: { Authorization: `Bearer ${loginRes.body.token}` }
})
console.assert(meRes.status === 200, 'Auth should work')
console.assert(meRes.body.email === user.email, 'Should return correct user')
// Cleanup
await db.users.delete({ id: user.id })
console.log('✓ Auth flow validated')
}
testAuthFlow().catch(console.error)
</scratch_directory>
<scenarios_jsonl>
Each line is complete JSON object with fields:
{
name: string // unique identifier (kebab-case)
description: string // human-readable summary
setup: string // prerequisites and state preparation
steps: string[] // ordered actions to execute
expected: string // success criteria
tags: string[] // categorization (auth, api, error, etc)
env?: string // required environment (staging, local, prod-readonly)
duration_ms?: number // typical execution time
}
Document in scenarios.jsonl when:
Delete from .scratch/ when:
</scenarios_jsonl>
Loop: Write → Execute → Document → Cleanup
Each iteration:
<gitignore_check>
MANDATORY before first .scratch/ use:
grep -q '.scratch/' .gitignore || echo '.scratch/' >> .gitignore
Verify .scratch/ directory will not be committed. All test scripts are local-only.
If .gitignore doesn't exist, create it:
[ -f .gitignore ] || touch .gitignore
grep -q '.scratch/' .gitignore || echo '.scratch/' >> .gitignore
</gitignore_check>
Prepare real dependencies:
Create .scratch/ test script:
Run proof program:
If scenario validates behavior:
Delete .scratch/ script or promote to permanent test suite.
ALWAYS:
NEVER:
ESCALATE when:
Patterns and examples:
Related skills:
External resources:
SOC 職業分類に基づく