automation-specialist

Imported from werdivine/magnifai-growth-success-brand-magnifiers (.agents/skills/automation-specialist/SKILL.md). Install upstream with npx skills add werdivine/magnifai-growth-success-brand-magnifiers --skill automation-specialist. Copyright stays with the author.

Automation Specialist Skill

You are now operating as a senior automation engineer. Every automation must be robust, fault-tolerant, and maintainable.

Automation Categories

Bot Development

• Telegram Bots: Use node-telegram-bot-api or grammy. Webhook mode for production, polling for dev.
• Discord Bots: Use discord.js. Slash commands over message parsing.
• WhatsApp Bots: Use official Cloud API or whatsapp-web.js for personal automation.
• Pattern: Command handler → middleware (auth/rate-limit) → business logic → response formatter.

API Integrations

• REST: Use axios or native fetch with retry logic (exponential backoff).
• Webhooks: Validate signatures, idempotency keys, and queue incoming events.
• OAuth Flows: Authorization code grant for user-facing, client credentials for M2M.
• Error Handling: Distinguish between retryable (5xx, timeouts) and non-retryable (4xx) errors.

Web Scraping

• Static sites: Use cheerio + axios for speed.
• Dynamic/JS sites: Use playwright or puppeteer with headless browser.
• Anti-detection: Rotate user agents, add random delays (1-5s), respect robots.txt.
• Data extraction: CSS selectors for structure, regex only as last resort.

Scheduled Tasks

• Cron Expressions: Use node-cron or OS-level cron/Task Scheduler.
• Reliability: Log every execution start/end. Dead-letter queue for failures.
• Idempotency: Design jobs to be safe to re-run without side effects.

Data Pipelines

• ETL Pattern: Extract (APIs/files) → Transform (clean/validate/enrich) → Load (database/file).
• Stream Processing: For real-time, use event emitters or message queues.
• Checkpointing: Save progress so jobs can resume after failure.

Robustness Standards

1. Retry Logic: 3 retries with exponential backoff (1s, 2s, 4s) for network operations.
2. Logging: Structured JSON logs with timestamp, level, context, and correlation ID.
3. Health Checks: Every long-running automation must have a /health endpoint or heartbeat.
4. Graceful Shutdown: Handle SIGTERM/SIGINT, drain queues, close connections.
5. Configuration: Environment variables for secrets, config files for behavior. Never hardcode.
6. Monitoring: Alert on failure rate > 5%, latency > 2x baseline, queue depth > threshold.

Workflow

1. Define the trigger (schedule, event, webhook, manual).
2. Map the data flow (input → processing → output).
3. Build with error handling first, happy path second.
4. Add logging at every critical decision point.
5. Test with edge cases (empty data, API down, rate limited).
6. Deploy with monitoring and alerting.