What problem does it solve? Raw product feedback recordings (screen capture, voice, and event logs) are hard to turn into actionable engineering input. This Skill analyzes Riffrec zip bundles, standalone video/audio, or meeting notes and produces structured evidence: transcripts, high-signal screenshots, candidate findings, and requirements-kickoff documents. ## Core Features & Use Cases - Three routing paths: setup guidance for installing Riffrec, a quick bug report path for short single-issue recordings, and an extensive analysis path producing a full Compound Engineering artifact set. - Automated evidence extraction: transcribes media via the OpenAI transcription API (with chunked fallback for long files), selects high-signal moments from click/network/console events, and extracts screenshots with ffmpeg. - Requirements handoff: generates analysis.md, problem-analysis.md, source-materials.md, and requirements-kickoff.md, then hands off to the ce-brainstorm skill for requirement confirmation. - Use Case: A user drops a riffrec-*.zip containing a 5-minute session where they verbalize a broken button. The Skill transcribes the voice track, extracts frames near the failed clicks, and produces a requirements document with timestamped evidence ready for brainstorming. ## Quick Start Analyze the attached riffrec-abc123.zip recording and turn it into structured product feedback.
