MoeenNehzati

Use when the user asks to install, update, propagate, or repair the `assistant`, `collab`, `coauthor`, or workspace helper commands, including missing or stale launchers and shell integration. Do not use for unrelated software or plugin installation.

Use when the user asks to set up or manage a recurring AI job. Do not use for one-off commands or generic scheduler questions.

Use when the user needs to set up or restore Google authentication for Famulus. Do not use for ordinary Google-service operations.

Use when the user asks to view or change a persistent personal list. Do not use for an ad hoc generated list, repository inventory, or prose checklist.

Use when the user or another skill needs to read from or write to the configured LLM root of a remote. Do not use for local files or remote paths outside that LLM root.

Use when the user asks to view or change their Google Calendar. Do not use for daily planning.

Use when the user asks to create a personal skill or change an existing personal skill's intended behavior or public interface in the shared skills directory. Do not use for behavior-preserving refactoring, blueprint regeneration, certificate work, or…

Use when the user asks for inbox-level email triage or processing. Do not use for ordinary email access, sending, or analysis of a single message.

Use when an arranged MusicXML score needs to be parsed into music_analysis.json — extracting tempo, meter, key, parts, phrases, melody candidates, and harmony — for downstream vocal planning. Does not read lyrics or alter the score.

Use when the user wants to transform or augment an existing symbolic score. Arranges MusicXML into a new styled score using preferences such as vibe, instruments, key, tempo, density, groove, bass, and drums while preserving score editability.

Use when sheet music or MIDI files need to be acquired for songs listed in a CSV catalog. Searches IMSLP, Musikverket, Internet Archive, MuseScore, and other sources. Saves downloads under assets/ with structured naming.

Use when a MIDI file needs to be rendered into audio. Prefers FluidSynth with a SoundFont; optionally converts WAV to compressed formats with ffmpeg.

Use when a vocal stem and an arranged score are ready to combine into a final mix. Renders the instrumental, aligns vocal timing, mixes levels, exports final_song.wav, and emits mix_report.json with timing, level, and musical conflict warnings.

Use when music_analysis.json and lyrics.json are ready and syllables need to be aligned to melody notes to produce vocal_events.json for singing synthesis. This is the main LLM-assisted musical decision step.

Use when a rough DiffSinger vocal is ready and you want to optionally improve its timbre and naturalness using an RVC voice conversion model before mixing. This is a refinement step, not the core singing step.

Use when sheet music in PDF or image format needs to be converted to MusicXML. Uses Audiveris for OMR recognition, saves outputs under assets/_build/xml/, and reports recognition weaknesses that need manual cleanup.