| name | feedback |
| description | Use when recording a user signal with its context, observation, implication, and confidence for future analysis. |
Feedback
Usage
Invoke as fiddle:feedback <feedback> — what the user said or experienced.
Append a structured user feedback entry.
Process
- Read the user's argument as the feedback content.
- If no argument, ask: "What did the user say or experience?"
- Ask for missing fields (one question at a time, skip fields the user already provided):
- Who — role, segment, experience level (e.g., "senior dev at fintech", "solo founder", "enterprise team")
- Context — where/when/how they encountered this (e.g., "during onboarding", "in Slack support channel", "user interview")
- If context is only available for some actors, use what you have and write "not stated" for the rest. Don't ask follow-ups for every actor.
- For each distinct actor or observation, append a structured entry to
docs/product/FEEDBACK.md.
Entry Format
### YYYY-MM-DD — <short observation title>
**Who:** <role, segment, experience level — or "internal" for team observations>
**Context:** <where/when/how this was observed>
**Observation:** <what happened — raw signal, no interpretation>
**Implication:** <what this suggests — one sentence>
**Confidence:** <high|medium|low>
Tags: #tag1 #tag2
Field Guidelines
- Who: Enough to identify the persona, not the person. "Senior dev at fintech" not "John from Acme Corp."
- Context: Channel and circumstance. "Slack support thread during first week of use."
- Observation: What happened — what they said, did, or struggled with. Facts only, no interpretation.
- Implication: What it means — one sentence on what this suggests for the product. This is where interpretation goes.
- Confidence:
high = direct quote or observed behavior. medium = secondhand report or partial observation. low = inference from indirect signal.
Multi-Actor Scenarios
When one report covers several actors ("three customers said X"), write one entry per actor, sharing the date and an observation title that names the common theme (for example, "Evaluator domain creation is opaque"). Each actor gets their own Who, Context, Observation, Implication, and Confidence. Flattening distinct experiences into a single entry loses the differences between actors, which is the part fiddle:insights reads to distinguish a persona-specific friction from a general one.
Tags
Assign 1-3 tags from: #feature-request #bug #confusion #praise #churn-signal #ux #performance #onboarding #docs.
Writing the Entry
Show the entry and append after confirmation, creating docs/product/FEEDBACK.md with the header # User Feedback if it does not exist. Keep Observation to 1-3 sentences and Implication to one.
The file is append-only: existing entries are never edited or deleted, because a synthesis run compares what was reported then against what is reported now.
