메뉴
Browser automation via ActionBook action manuals. Use for web interactions, deployment verification, visual testing, and admin UI tasks. 10x faster than raw browser automation with 100x token savings.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
| name | actionbook |
| description | Browser automation via ActionBook action manuals. Use for web interactions, deployment verification, visual testing, and admin UI tasks. 10x faster than raw browser automation with 100x token savings. |
| triggers | ["browser","website","screenshot","deploy check","visual test","open page","actionbook","action manual","web interaction"] |
| version | 1 |
| last_updated | "2026-02-01T00:00:00.000Z" |
ActionBook provides action manuals - pre-built, reusable browser automation workflows that are 10x faster and use 100x fewer tokens than raw Playwright automation.
Use ActionBook when you need to interact with websites in predictable, repeatable ways:
Key benefit: Action manuals are pre-tested, optimized workflows that work reliably without trial-and-error debugging.
ActionBook provides 3 MCP tools:
| Tool | Purpose | Example |
|---|---|---|
searchActions | Find action manuals by keyword | Search for "login", "checkout", "form" |
getActionById | Get specific action manual details | Retrieve manual by ID |
list_sources | List available action sources | See all available manuals |
CRITICAL: You MUST use ToolSearch("actionbook") BEFORE calling these tools to load them.
// 1. Load ActionBook tools (REQUIRED first step)
ToolSearch("actionbook");
// 2. Search for relevant action manual
searchActions({ query: "deployment verification" });
// 3. Get specific manual details
getActionById({ id: "verify-app-live" });
// 4. Execute the action manual
// (Follow the manual's instructions)
Priority: ActionBook (if manual exists) > Playwright (custom E2E) > Claude-in-Chrome (interactive)
After deploying to Accept or Production:
// Search for deployment verification manual
searchActions({ query: "deployment check health" });
// Execute manual to verify:
// - App is accessible
// - Login works
// - Critical pages load
// - No console errors
Value: Automated post-deploy smoke test without writing custom E2E tests.
Capture screenshots for design comparison:
// Search for screenshot manual
searchActions({ query: "screenshot responsive" });
// Execute manual to capture:
// - Desktop (1920x1080)
// - Tablet (768x1024)
// - Mobile (375x667)
Value: Consistent screenshot capture across breakpoints for design reviews.
Interact with admin panels:
// Search for admin interaction manual
searchActions({ query: "admin panel crud" });
// Execute manual for:
// - Prisma Studio data inspection
// - Admin panel configuration
// - User management tasks
Value: Automate repetitive admin tasks without writing custom scripts.
Setup tracking and marketing tools:
// Search for analytics setup manual
searchActions({ query: "google analytics setup" });
// Execute manual for:
// - Analytics code verification
// - Event tracking validation
// - Social media integration checks
Value: Verify marketing integrations without manual clicking.
Scenario: Verify landing page after deployment
// Step 1: Load ActionBook tools
ToolSearch("actionbook");
// Step 2: Search for relevant manual
const results = searchActions({
query: "landing page verification screenshot",
});
// Step 3: Review results and select manual
// Results show: "Landing Page Health Check" (id: "lp-health-001")
// Step 4: Get full manual details
const manual = getActionById({ id: "lp-health-001" });
// Step 5: Follow manual instructions
// Manual includes:
// - Navigate to URL
// - Check page load time
// - Verify hero section visible
// - Capture screenshot
// - Check console for errors
// - Verify CTA buttons clickable
// Step 6: Execute and report results
Time saved: 5-10 minutes vs writing custom Playwright test
Tokens saved: ~95% vs raw browser automation
// Generic search
searchActions({ query: "login" });
// Specific search
searchActions({ query: "google oauth login" });
// Multi-keyword
searchActions({ query: "form validation error" });
// 1. Find verification manual
searchActions({ query: "app health check" });
// 2. Find screenshot manual
searchActions({ query: "responsive screenshot" });
// 3. Execute in sequence:
// - Verify app is healthy
// - Capture screenshots for review
// Before deployment
searchActions({ query: "screenshot capture" });
// Execute manual, save screenshots to /before/
// After deployment
// Execute same manual, save screenshots to /after/
// Compare with visual diff tool
ActionBook complements our existing browser automation:
| Tool | Purpose | Location |
|---|---|---|
| ActionBook | Pre-built workflows | MCP server (global) |
| Playwright E2E | Custom test suites | /e2e-tests/ |
| Claude-in-Chrome | Interactive browsing | MCP server (global) |
Workflow recommendation:
| Task | ActionBook Manual | Alternative |
|---|---|---|
| Post-deploy verification | "deployment health check" | Custom Playwright test |
| Screenshot capture | "responsive screenshot" | Manual browser + DevTools |
| Form testing | "form validation flow" | Custom Playwright test |
| Login flows | "oauth login" | Custom auth test |
| Analytics verification | "analytics tracking check" | Manual verification |
list_sources to see all available manualsnpm i -g @actionbookdev/mcp.mcp.json configuration@actionbookdev/mcp.mcp.jsontroubleshooting-guide, playwright (via MCP)/e2e-tests/CLAUDE.mdActionBook vs Raw Automation:
| Approach | Tokens | Time | Reliability |
|---|---|---|---|
| ActionBook manual | ~500 | 1-2 min | High (pre-tested) |
| Raw Playwright | ~50,000 | 10-30 min | Medium (trial-error) |
| Claude-in-Chrome | ~20,000 | 5-15 min | Medium (exploratory) |
Savings: 100x tokens, 10x faster, higher reliability
When ActionBook shines:
When to skip ActionBook:
Complete code quality workflow with ESLint, Prettier, TypeScript, Husky hooks, and type safety standards. Run before every commit/push.
Comprehensive code review skill for Wasp applications. Use when reviewing code, PRs, or architectural changes. Covers Wasp-specific patterns, import rules, auth checks, TDD compliance, multi-tenant permissions, and performance. Integrates all CLAUDE.md critical rules.
Complete error handling patterns for Wasp applications. Use when implementing error handling, validation, or working with HTTP errors. Includes server-side operations errors, client-side error handling, Zod validation, and retry logic.
Type safety standards and linting philosophy for TypeScript/Wasp projects. 2-tier approach (production strict, tests pragmatic). Helper patterns, mock patterns, when 'any' is acceptable vs cheating.
Multi-tenant permission checking for Wasp applications. Use when implementing authorization, access control, or role-based permissions. Includes organization/department/role patterns and permission helper functions.
Execute GREEN phase of TDD for Ralph automation. Minimal implementation to make tests pass with gate tracking. Use via Skill tool in Ralph loops.