matthiasn

Standard for how character-animation PRs present visual evidence: before/after render grids, true-speed GIFs, movement-curve comparisons, and panel scorecards embedded directly in the PR body/comments. Use whenever opening or commenting on a PR that touches a…

Generate the review artifacts for character animation: contact-sheet grids, per-frame sequences, true-speed animated GIFs, motion-trace (movement-curve) sheets, onion skins, transition strips, and before/after comparisons. Use whenever renders, GIFs, curves,…

Export the beat-synced character dance showcase to a verified MP4 with synced AAC audio using the repo's realtime Xwayland exporter, release-app exact-frame exporter, or Flutter-test fallback and ffmpeg. Use when the user wants a 1920x1080 or 1280x720 video…

Generate real mouth-shape lip-sync cues for the character dance demo from a song's vocal stem, using Rhubarb Lip Sync via tools/dance_audio/lipsync.py. Use when the user wants believable lip movement (not one-vowel-per-word), to "lip sync the cats", produce a…

Prepare any song for the beat-synced character dance demo — produce its beat/downbeat map and word-timestamped lyrics (optionally lyric-forced-aligned with lead/background tags), using the offline tools in tools/dance_audio. Use when the user wants to "add a…

Prepare generated or hand-painted scenery art for layered Flutter backdrops: full-frame PNG plates, alpha masks, cloud/wave isolation, OpenCV cleanup, runtime stacking, parallax validation, and visual QA. Use when Codex needs to split a base backdrop into…

Verify, regenerate, and debug full-frame scenery masks and alpha layers against a single master plate. Use when derived backdrop layers, masks, moving clouds, aircraft, drones, reflections, or other animated scenery appear shifted, clipped, ghosted,…

Run a grounded expert panel on character animation using real rendered frame strips, dense frame captures, or video stills. Use when motion looks robotic, lame, snappy, wobbly, drunk, physically impossible, off-beat, weakly choreographed, or when the user…

Build, debug, extend, localize, or publish the automated tutorial videos — the workbench that drives the real Linux app under Xvfb (virtual mic, live Voxtral transcription, task-agent proposals), records the screen, narrates with Gemini or ElevenLabs TTS,…

Analyze Drift / SQLite slow-query and super-slow-query logs against this app's known stall patterns (read waves, N+1, transaction scoping, MultiExecutor contention, WAL/OS factors)

Capture in-app screenshots of a widget or flow at mobile + desktop sizes — offline, deterministic, real fonts and icons — using the reusable screenshot harness. Use when asked to "show me X in the app", "screenshot this screen", or to preview a UI change for…

Run a multi-agent design review on a UI surface — capture reproducible baseline screenshots, then rate them with a panel of design experts (one agent per craft dimension) and, optionally, a panel of user personas with different cognitive styles. Iterate…

Fetch PR review comments, address each one in code, and post resolution replies

Act as a skeptical senior engineer performing a detailed code review of the latest changes on the current branch (or a given PR) — best practices, maintainability, performance, security, and readability. Objective and concise; only real issues, with a clear…

Cut a Lotti release — assemble the changelog.d/ fragments into CHANGELOG.md and the Flathub metainfo, bump the version, open the release PR, then tag. Use when asked to cut or prepare a release, assemble the changelog, publish a new version, or work out what…

Use for authorized Lotti maintainer work that needs private durable task tracking, issue dependencies, blocker management, multi-session handoff, or shared agent memory. Public contributors use GitHub Issues instead.