ams-workflow-automation
暂无简介
- License 许可
- MIT
- 最近更新
- 2026-11-07
- Stars
- ★ 0
- Forks
- 1
03 SKILL.md 正文
Workflow Automation
Overview
Workflow automation connects triggers to actions: when X happens, do Y — across apps, APIs, and systems.
Reliable automation is a design discipline: idempotent actions, explicit error handling, retries with backoff, and observability from day one.
This skill is platform-agnostic: the principles apply whether you use n8n, Zapier, Make, code, or cron jobs.
The question is never 'can I automate this?' but 'should I — and how will I know when it breaks?'
When to use
- Repetitive multi-step processes done manually
- Connecting apps that don't integrate natively
- Evaluating automation platforms for a team
- Automations that keep breaking silently
- Deciding what to automate vs. leave manual
Core concepts
- Triggers. Events that start the flow: webhooks (instant), polling (scheduled checks), schedules (cron), manual. Prefer webhooks over polling where available — faster and cheaper.
- Actions and branching. Steps that do things, with conditional branches and loops. Keep flows linear where possible; deep branching becomes untestable.
- Idempotency. Running the same flow twice shouldn't duplicate effects. Design actions to be safe on retry: check-before-create, upsert instead of insert, dedupe keys.
- Error handling. Every external call can fail. Retry with exponential backoff, dead-letter queues for poison messages, and alerts — not silent drops.
- State and dedup. Track processed items (IDs, timestamps) to avoid reprocessing. Polling triggers especially need high-water marks.
- Observability. Log runs, alert on failures, dashboard key metrics. An automation you can't monitor is a liability with a timer.
- Secrets management. API keys in vaults/secrets managers, never hardcoded. Rotate credentials; scope them to least privilege.
- Manual escape hatches. Every automation needs a pause button, a manual retry, and a way to run steps by hand. You'll need them at 2am.
Practical workflow
- Map the process manually. Document the current manual process step by step. Automate the stable version — automating a messy process cements the mess.
- Score candidates. Frequency x time-cost x error-cost. Automate high-frequency, error-prone drudgery first. Leave rare, judgment-heavy work manual.
- Choose the platform. No-code (fast, limited) vs. code (flexible, maintained) vs. hybrid. Match complexity: simple linear flows -> no-code; branching logic -> code.
- Design for failure. For each step ask: what if the API is down? What if data is malformed? What if it runs twice? Build retries, validation, and dedup in.
- Build the happy path. Get the core flow working end-to-end before edge cases. Test with real data in a sandbox.
- Add observability. Logging, failure alerts (to a channel you actually watch), run dashboards. Test the alerting by breaking something deliberately.
- Document and hand off. What it does, how to pause/retry, who owns it. Automations outlive their builders — document like you'll forget.
- Review periodically. APIs change, processes evolve. Quarterly: still needed? Still working? Still the right tool? Prune dead automations.
Common pitfalls
- Automating a broken process. Automating chaos produces faster chaos. Fix the process manually first, then automate the stable version.
- No idempotency. Double-running creates duplicate orders, emails, tickets. Design every action safe-on-retry from the start.
- Silent failures. Automation breaks, nobody notices for weeks. Alerts on failure are not optional — they're the product.
- Polling everything. 5-minute polling across 20 integrations burns quota and lags. Webhooks where available; sensible intervals elsewhere.
- Hardcoded secrets. API keys in flow definitions, shared screenshots, repos. Vaults + rotation, always.
- No pause button. Runaway automation spamming customers with no kill switch. Every flow needs an off switch you can hit in seconds.
- Over-automation. Automating rare, judgment-heavy tasks because you can. Some work should stay human — automate the rote, not the judgment.
- Orphaned automations. Builder leaves, automation breaks, nobody knows it exists. Ownership + documentation + monitoring, or delete it.
04 同仓库更多 Skills
web-scrape
Turn web pages into LLM-ready markdown via the Firecrawl API (renders JS, anti-bot handling). Use when browser_open/fetch fails on JS-heavy pages, or when you need clean markdown/structured extraction from a site. Requires FIRECRAWL_API_KEY.
design-taste-frontend
tasteskill v2, the anti-slop frontend skill. Art direction for websites, distilled from a study of 247 curated gallery sites (One Page Love, Awwwards Sites of the Day). Deals researched directions (idea, composition, type, palette, photographs, section blueprints) from a library, enforces a strict no-slop house style with a real-browser inspector, and runs a look-and-fix loop on screenshots. Use this whenever you build, design, redesign, restyle, polish or review a landing page, marketing site, portfolio, product page, one-pager, hero or any public-facing web page, in any stack (HTML/CSS, React, Next.js, Vue, Svelte, Astro, Tailwind) — also when the user says a page looks generic, templated, "AI-made", sloppy, or asks to make it look designed, premium, award-level or less like every other site. Not for dashboards or app screens.
test-driven-development
Use when implementing any feature or bugfix, before writing implementation code
systematic-debugging
Use when encountering any bug, test failure, or unexpected behavior, before proposing fixes
requesting-code-review
Use when completing tasks, implementing major features, or before merging to verify work meets requirements
brainstorming
You MUST use this before any creative work - creating features, building components, adding functionality, modifying behavior, or planning anything new, in software or out of it (a talk, a business, a renovation).