produck-feedback

Name: produck-feedback
Author: tryproduck

487stars7forks

How to install produck-feedback?

tryproduck/produck-skills/produck-feedback

$npx -y skills add tryproduck/produck-skills --skill produck-feedback

Installs into the current project.

Summary

Use when acting on real user feedback — you have a Produck feedback ticket or URL, or the user wants to fix something users complained about or build a feature users asked for. Pulls full feedback context through the Produck MCP and drives a feedback → aligned PRD → build loop. T

Files · 1

View on GitHub

SKILL.md

1	# Building from Produck feedback
2
3	Purpose: Turn real, in-context user feedback into shipped fixes. This skill drives the
4	[Produck](https://tryproduck.com) MCP — it tells the agent how to find the right feedback, load the
5	full context behind it, and convert that context into an aligned change instead of a guess.
6
7	> Requires the Produck MCP server connected. If `search_feedback` / `get_feedback` are not available,
8	> see [`references/connect-mcp.md`](references/connect-mcp.md) to connect it and mint a token first.
9
10	---
11
12	## 1. What this connects to
13
14	Produck captures feedback in the user's own session, so a ticket is far richer than a text
15	complaint. Each ticket can carry:
16
17	- Written and spoken feedback — the user's typed note plus any voice transcription.
18	- Reconstructed page HTML — a static snapshot of the exact page the user was looking at.
19	- Annotation markers — the specific elements/regions the user pointed at.
20	- Session activity — a timeline of what the user did leading up to the feedback.
21	- A generated design doc — Produck's own first-pass analysis of the issue, when present.
22
23	The job of this skill is to pull that context and use all of it, not just the title.
24
25	## 2. The two MCP tools
26
27	The server is named `produck` (tools may appear prefixed, e.g. `mcp__produck__search_feedback`).
28
29	### `search_feedback` — find relevant tickets
30	Returns lightweight summaries; use it to locate tickets, then load full context with `get_feedback`.
31
32	\| Input \| Type \| Notes \|
33	\| --- \| --- \| --- \|
34	\| `domain` \| string, optional \| Filter to one website domain. \|
35	\| `limit` \| int 1–50, optional \| Defaults to 20. \|
36	\| `cursor` \| string, optional \| Pagination cursor from a previous call. \|
37
38	Returns `{ items, nextCursor, count }` — each item a summary (id, domain, page URL, created-at,
39	annotation count, source). Page with `cursor` until `nextCursor` is null when you need the full set.
40
41	### `get_feedback` — load one ticket in full
42	\| Input \| Type \| Notes \|
43	\| --- \| --- \| --- \|
44	\| `feedbackId` \| string, required \| The ticket id (from a `search_feedback` item, a URL, or the user). \|
45
46	Returns the full ticket: written/spoken feedback, reconstructed HTML, annotations, session activity,
47	and the design doc when present. Always `get_feedback` before building — never act on a summary alone.
48
49	## 3. The operating loop
50
51	```text
52	search_feedback → find the relevant ticket(s) (filter by domain, page as needed)
53	get_feedback → load ONE ticket's full context
54	user-alignment → turn the ticket into an agent-executable PRD (do not skip)
55	build → implement the fix / high-fidelity mockup
56	verify → check the change against the ticket's annotations + design doc
57	```
58
59	- Do not jump from a raw ticket straight to code. A ticket is a user request, not a spec. Hand
60	it to the `user-alignment` skill first to produce an aligned PRD — interpreted intent, scope,
61	non-goals, acceptance criteria — then build from that. The two skills are designed to compose.
62	- Ground every decision in the captured context. Use the reconstructed HTML and annotation markers
63	to locate the exact elements involved; use the session activity to understand what the user was
64	doing; use the design doc as a starting hypothesis, not the final answer.
65	- Close the loop. When you finish, verify the result against the annotations and the stated
66	outcome — the same ticket is your acceptance check.
67
68	## 4. Setup
69
70	To connect the MCP, choose the install path for your agent and mint a token — see
71	[`references/connect-mcp.md`](references/connect-mcp.md).

People who liked this also viewed

6 picks

Type

produck-feedback

487stars7forks

View on GitHub

How to install produck-feedback?

tryproduck/produck-skills/produck-feedback

$npx -y skills add tryproduck/produck-skills --skill produck-feedback

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# Building from Produck feedback
2
3	Purpose: Turn real, in-context user feedback into shipped fixes. This skill drives the
4	[Produck](https://tryproduck.com) MCP — it tells the agent how to find the right feedback, load the
5	full context behind it, and convert that context into an aligned change instead of a guess.
6
7	> Requires the Produck MCP server connected. If `search_feedback` / `get_feedback` are not available,
8	> see [`references/connect-mcp.md`](references/connect-mcp.md) to connect it and mint a token first.
9
10	---
11
12	## 1. What this connects to
13
14	Produck captures feedback in the user's own session, so a ticket is far richer than a text
15	complaint. Each ticket can carry:
16
17	- Written and spoken feedback — the user's typed note plus any voice transcription.
18	- Reconstructed page HTML — a static snapshot of the exact page the user was looking at.
19	- Annotation markers — the specific elements/regions the user pointed at.
20	- Session activity — a timeline of what the user did leading up to the feedback.
21	- A generated design doc — Produck's own first-pass analysis of the issue, when present.
22
23	The job of this skill is to pull that context and use all of it, not just the title.
24
25	## 2. The two MCP tools
26
27	The server is named `produck` (tools may appear prefixed, e.g. `mcp__produck__search_feedback`).
28
29	### `search_feedback` — find relevant tickets
30	Returns lightweight summaries; use it to locate tickets, then load full context with `get_feedback`.
31
32	\| Input \| Type \| Notes \|
33	\| --- \| --- \| --- \|
34	\| `domain` \| string, optional \| Filter to one website domain. \|
35	\| `limit` \| int 1–50, optional \| Defaults to 20. \|
36	\| `cursor` \| string, optional \| Pagination cursor from a previous call. \|
37
38	Returns `{ items, nextCursor, count }` — each item a summary (id, domain, page URL, created-at,
39	annotation count, source). Page with `cursor` until `nextCursor` is null when you need the full set.
40
41	### `get_feedback` — load one ticket in full
42	\| Input \| Type \| Notes \|
43	\| --- \| --- \| --- \|
44	\| `feedbackId` \| string, required \| The ticket id (from a `search_feedback` item, a URL, or the user). \|
45
46	Returns the full ticket: written/spoken feedback, reconstructed HTML, annotations, session activity,
47	and the design doc when present. Always `get_feedback` before building — never act on a summary alone.
48
49	## 3. The operating loop
50
51	```text
52	search_feedback → find the relevant ticket(s) (filter by domain, page as needed)
53	get_feedback → load ONE ticket's full context
54	user-alignment → turn the ticket into an agent-executable PRD (do not skip)
55	build → implement the fix / high-fidelity mockup
56	verify → check the change against the ticket's annotations + design doc
57	```
58
59	- Do not jump from a raw ticket straight to code. A ticket is a user request, not a spec. Hand
60	it to the `user-alignment` skill first to produce an aligned PRD — interpreted intent, scope,
61	non-goals, acceptance criteria — then build from that. The two skills are designed to compose.
62	- Ground every decision in the captured context. Use the reconstructed HTML and annotation markers
63	to locate the exact elements involved; use the session activity to understand what the user was
64	doing; use the design doc as a starting hypothesis, not the final answer.
65	- Close the loop. When you finish, verify the result against the annotations and the stated
66	outcome — the same ticket is your acceptance check.
67
68	## 4. Setup
69
70	To connect the MCP, choose the install path for your agent and mint a token — see
71	[`references/connect-mcp.md`](references/connect-mcp.md).

People who liked this also viewed

6 picks

Type