The n8n template that turns a form response into a cited spec
A downloadable n8n workflow: form or webhook in, evidence-backed insight and PRD problem statement out, posted to Slack. The Specky call needs no API key, so it produces a real result the first time you import it.
- Steps
- 5
- Tools
- 6
- Metrics
- 3
- Read
- 4 min
An automation engineer with an n8n instance and a client who keeps saying “we get a lot of feedback and we never do anything with it.”
Composite blueprint. This is the loop and the wiring — assembled from workflows our early design partners run — not a single named customer’s audited results. Every number on this page is a metric to instrument, not a result we’re claiming.
What the week looked like
The client had a Tally form, a support inbox, and a Slack channel where feedback went to die.
Every “let’s finally do something with this” project stalled in the same place: an LLM node with a summarise-this prompt, producing summaries that told nobody anything they didn’t already know.
Summaries aren’t decisions. The client wanted to know what to build, and a prompt inside a workflow can’t answer that — it has no memory of anything else that customer ever said.
Building the memory layer yourself is a vector store, an ingestion pipeline, and a maintenance burden no retainer covers.
One tool call replaces the prompt node. It returns a structured insight and a PRD problem statement with the verbatim quotes attached — and it does it without an account, so you can run it live in the discovery call and wire it properly afterwards.
The tools this loop runs on
Native connectors where they exist, and an honest label where they don't — anything else reaches Specky over MCP, the REST API, or an automation hop.
MCP Client Tool node pointed at Specky, or a plain HTTP Request node if you’d rather not add one. The template ships both paths.
Specky’s keyless tool. No API key, no signup, IP rate-limited at 10/hour. This is what makes the template runnable on import.
Whatever already collects the feedback. Swap the trigger node; nothing downstream changes.
Where the result lands, formatted as a decision rather than a digest.
Optional branch: when the insight clears a confidence threshold, open a ticket.
The upgrade path. Add a workspace key and the same workflow starts writing into a graph that remembers — clustering across months instead of judging each response alone.
How it actually runs
Every step is labelled by who does it. The steps marked 'you review' are the ones worth keeping human.
Import the template
AutomaticDownload the JSON, import it into n8n, hit execute. It works immediately, because the Specky node needs no credential. Verifying that a template does something useful before you invest in it is the whole reason this one is keyless.
Point it at whatever collects feedback
AutomaticReplace the trigger with your client’s Tally form, Typeform, support inbox, or a bare webhook. One node changes; the rest of the workflow doesn’t care where the text came from.
The tool call replaces the prompt node
AgentInstead of asking a model to summarise, the workflow calls feedback_to_spec and gets back a typed result: the insight, the problem statement, and the verbatim quotes that support it. Structured output means the next node can branch on it.
Slack gets a decision, not a digest
You reviewThe post says what the user is actually trying to do, what it would take, and which sentence justified that reading. A digest gets skimmed; a claim with a quote under it gets replied to.
Add a key when it stops being a demo
You reviewDrop a workspace key into the node and the same workflow starts writing into a graph. That’s when one response stops being judged alone and starts being compared against everything else the client’s users have said.
The numbers to watch — not the ones we'd like to quote
We're not going to invent a percentage for you. These are the three metrics that tell you whether the loop is real, and how to compute each one from data the graph already holds.
Time from import to first useful output
Under five minutes, from a cold n8n instance
Actually time it on a fresh import. If a template needs an account before it does anything, it doesn’t get evaluated — it gets closed.
Share of forwarded feedback that produces a decision
Higher than the summarise-this baseline it replaced
Count Slack posts that got a human reply or became a ticket, over total posts. Run the old prompt node for a week alongside it if you want the honest comparison.
Whether the client asks about the graph
Within the first month
Not a dashboard metric — a conversation. The keyless version deliberately can’t cluster across time, so the first “has anyone else said this?” is the signal the workspace is worth what it costs.
Start here, not everywhere
Download the template
Grab the workflow JSON above and import it into n8n. It uses a plain HTTP Request node so it runs on any n8n version, with a note in the canvas showing the MCP Client Tool alternative if your build has that node.
Swap the trigger, leave the rest
The Specky node takes one string. Anything that produces text — a form field, an email body, a Slack message, a review — works without touching the downstream nodes.
Add the key when the client is real
Keyless is for evaluation and demos. The moment the client wants “has anyone else said this?”, they need a workspace — and that’s the conversation the template exists to earn.
https://www.specky.space/api/mcpClaude Code · Codex · Cursor · Claude Desktop · ChatGPT · n8n
Frequently asked questions
Do I need a Specky account to run this template?+
No. feedback_to_spec is callable with no Authorization header at all, rate-limited by IP at 10 calls an hour. Import the template, execute it, and you get a real structured result — that’s deliberate, because a template that demands a signup before it demonstrates anything doesn’t get evaluated.
What’s the catch on the keyless tool?+
It judges each piece of feedback alone. It can’t tell you nine other accounts said the same thing in March, because without a workspace there’s no graph to compare against — and that comparison is the actual product. The keyless tool is an honest demo of one step in a longer loop, not a free tier of the whole thing.
Can I use this in client work or resell it?+
Yes, and there’s a partner programme for exactly that — see /partners. If you’re building this into retainers, talk to us before you price it; the workspace-per-client shape has implications for how you bill that are worth five minutes of conversation.
Run this loop on your own product.
One source, one agent, one approval gate. You’ll know inside a week whether it holds.
14-day free trial · No credit card required · Cancel anytime