feedbacklisted
Install: claude install-skill yusufkaracaburun/ai-kit
# Send feedback to ai-kit
Capture one structured piece of feedback from the user about ai-kit — friction, surprise, clarity, gap, or win — and open a GitHub issue against `yusufkaracaburun/ai-kit` using the **feedback** issue template. This is how ai-kit improves at < 50 users: real friction, surfaced cheaply, triaged in the open.
## Not this skill
- **Bug in ai-kit** — use the `bug` template directly via `gh issue create --repo yusufkaracaburun/ai-kit -t bug`. Bugs are reproducible defects; feedback is everything else.
- **Propose a new skill** — use the `skill-suggestion` template. Specific design proposal, not "I wish there was …".
- **Question or open-ended discussion** — point the user at https://github.com/yusufkaracaburun/ai-kit/discussions instead of opening an issue.
- **Bug in the user's own project** — out of scope. `/ai:feedback` is feedback about ai-kit, not the project being built with it.
If the user's input clearly matches one of the above, route them and stop.
## Process
### 1. Detect the kind
Ask the user one question to pick one of:
- **Friction** — the skill/rule worked but felt clunky (too many turns, too much output, weird default).
- **Surprise** — behaviour differed from what the description led the user to expect.
- **Clarity** — docs / skill description / output was unclear.
- **Gap** — a real situation isn't covered by any current skill/rule.
- **Win** — what worked well (file this so we don't regress it).
Skip the question if the user alre