Explore templates as learning shortcuts
- Fork a Dify template and an n8n template; change exactly one thing in each
- Read a well-built template to extract a reusable pattern
- Treat "start from template" as the default, "from scratch" as the exception
Fork, don't start from scratch. The fastest way to ship — and to learn what good looks like — is to clone a working template, read how it's built, and change one thing. Students who always start from scratch spend 80% of their time re-inventing things they could have inherited.
- If you fork a template and change 10 things before running it once, how do you know which change caused a failure?
- Why is reading a template often more educational than reading docs?
- **Dify's Explore tab** has 30+ cloneable apps (chatbots, agents, workflows) covering RAG, research, coding, writing. Each is a complete app you can duplicate into your workspace in one click and then edit. - **n8n's Template Library** (menu → Templates, or https://n8n.io/workflows) has thousands of community workflows covering Slack bots, webhooks, AI automations, data sync.
Reading a well-built template teaches more than reading documentation, because you see real node choices and real prompt design in context.
Instead of designing a paper-chat RAG app from scratch, clone Dify's "Knowledge Q&A" explore app, swap in your own Knowledge Base, tweak the system prompt, and ship in 10 minutes. Instead of wiring a GitHub-to-Slack notifier by hand, fork an n8n template and change two credentials.
▸ Use the instructor's finished build before you build yours feel what "done" looks like — then recreate it
Not loading? https://dify.32dots.de/chat/YSqFa8QI6TB6ooKU
Go to **Studio → Explore** (or the Explore tab at the top). Pick the template closest to your capstone project. Click **Add to Workspace**. Open it, change exactly ONE thing (prompt, model, or knowledge base), ship it, and write down what you changed and why. ### n8n task Go to **n8n → Templates** (or https://n8n.io/workflows). Filter by "AI" or "Research". Import a template you find interesting. Run it. Change exactly ONE thing (credential, trigger, or an LLM node's model), run it again, and write down the diff. ### Blocks needed - Dify: Explore → clone → edit one thing → publish - n8n: Templates → import → run → edit one thing → run - No custom nodes required
Dify: Explore → clone → edit one thing → publishn8n: Templates → import → run → edit one thing → runNo custom nodes required
Go to **n8n → Templates** (or https://n8n.io/workflows). Filter by "AI" or "Research". Import a template you find interesting. Run it. Change exactly ONE thing (credential, trigger, or an LLM node's model), run it again, and write down the diff. ### Blocks needed - Dify: Explore → clone → edit one thing → publish - n8n: Templates → import → run → edit one thing → run - No custom nodes required
Dify: Explore → clone → edit one thing → publishn8n: Templates → import → run → edit one thing → runNo custom nodes required
Both tools explicitly design for forking — this card is about using that. Dify templates teach cognition patterns (prompt shape, tool selection); n8n templates teach plumbing patterns (trigger → transform → route). You want both vocabularies.
- **Changing too much at once** — "I'll just tidy it up while I'm in there" breaks the baseline. Change one thing, ship, then iterate.
- **Forking without reading** — if you can't explain why the original chose this topology, you haven't learned anything from it.
- **Dismissing simple templates** — a 3-node template that solves a real problem is a better teacher than a 30-node "ultimate" workflow.
Why is cloning a template and modifying one thing often a better first project than building from scratch?
Fork one Dify template + one n8n template relevant to your capstone. Change one thing in each. Ship both.
Two forked-and-modified apps/workflows + a 1-paragraph note per fork: what was the original for, what did you change, what did you learn.
Of the templates you looked at, which one would you steal a *pattern* from — not the whole thing, just one design choice — for your capstone?