prototype-feedback-loop

Distills founder video walkthroughs of prototypes into structured docs and frame-attached tickets.

1|Updated Jun 30, 2026
One-click install
npx skills add https://github.com/Lia-Creative/lia-plugins --skill prototype-feedback-loop-lia-creative
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: prototype-feedback-loop
Source: https://github.com/Lia-Creative/lia-plugins/tree/main/lia-tools/skills/prototype-feedback-loop
Command: npx skills add https://github.com/Lia-Creative/lia-plugins --skill prototype-feedback-loop-lia-creative

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires ffmpeg, openai-whisper.

What problem does it solve? Founders review builds by narrating screen-by-screen video walkthroughs, but raw video is not searchable, cannot be cited from a ticket, and does not separate items that can be built straight from items that need a jam session first. This Skill turns that fast capture format into a structured, linkable record the whole team can act on. ## Core Features & Use Cases - Stage-appropriate feedback gate: Detects structure-stage artefacts (lo-fi wireframes, stripped-back designs) and blocks detail-rubric scoring and premature ticket creation, returning only shape-level contradictions instead. - Transcript distillation: Converts a plain-text transcript and timestamped .srt file into a product record with the founder's overall read, section-by-section notes, chapter marks, and a split between build-straight items and needs-a-jam items. - Video-first ticket pipeline: Transcribes raw video locally with ffmpeg and openai-whisper, then creates one ticket per surface with the exact video frame attached via the Linear MCP upload flow. - Use Case: A founder records a 30-minute reaction to the current build. The Skill transcribes it, files a dated feedback doc with chapter marks, and produces tickets whose descriptions quote the founder's words, cite timestamps, and attach the relevant frame. ## Quick Start Distill this founder feedback video into a structured record and create frame-attached tickets for each item that needs work.

Frequently Asked Questions about prototype-feedback-loop

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I turn a founder feedback video into tickets?▼

Transcribe the video locally with ffmpeg and openai-whisper using the small.en model, then distil the transcript into sections and create one ticket per surface or theme. Each ticket quotes the founder's words, cites the video timestamp, and attaches the frame extracted at that moment.

How to extract a frame from a video at a specific timestamp?▼

Use ffmpeg with the -ss flag for the timestamp and -frames:v 1 to grab a single frame, for example ffmpeg -ss 00:12:34 -i recording.mp4 -frames:v 1 frame.jpg. Find the timestamp by grepping the .srt transcript for the item's opening phrase.

What feedback should I give on a structure-stage wireframe?▼

Report only where the shape contradicts itself, where one element does two jobs, and whether the chosen limitation holds. Do not score it against a detail rubric, do not create tickets, and put missing-detail observations in a clearly marked parking lot or omit them.

Why is the .srt transcript file required for feedback distillation?▼

The .srt file carries the timestamps that make chapter marks and frame extraction possible. Without it you cannot locate where each section starts in the video, so ask for it whenever only a plain-text transcript arrives.

When should feedback not become build tickets?▼

When the artefact is structure-stage, tickets create backlog pressure on an unsettled shape and are blocked by the stage gate. Also, a founder asking for a brainstorm about a topic gets a jam ticket labeled Research, explicitly not a build ticket.