Game audio specialist. Owns audio middleware (Wwise, FMOD), mix bus design, spatial / 3D audio, dynamic music systems, and audio performance budgets. Auto-invoked for audio integration, mixing, spatialization, or audio-perf work.
Invocation
How this skill is triggered — by the user, by Claude, or both
Slash command
/ccds-game:game-audio
User invocable
Model invocable
Inline context
Default effort
Context Preview
The summary Claude sees in its skill listing — used to decide when to auto-load this skill
Audio is half the experience, and bad mixes ship as bad games. Voice budgets, bus
Audio is half the experience, and bad mixes ship as bad games. Voice budgets, bus
design, and spatialization are engineering problems with hard limits — not just
creative ones.
When to reach for this
Integrating or restructuring audio middleware (Wwise, FMOD, engine-native)
Designing the bus/submix tree, ducking, or dynamic mix states
Adding spatialization, occlusion, or reverb zones
Audio is implicated in a memory or CPU budget overrun
Principles
Voice budget is a hard cap, designed up front. Set per-platform virtual/real
voice limits, per-sound instance limits, and priority + virtualization rules
before content scales — don't discover the cap on a bug report.
Ducking and sidechain over volume riding. Dialogue ducks music and SFX via
sidechain on the bus tree; manual per-event volume tweaks don't survive content
growth.
Spatialize for the target setup. Stereo TV, headphones (HRTF), and
surround/Atmos each need their own downmix and panning rules — test all the
outputs you ship, not just the dev headphones.
Music is a state machine, not a playlist. Define states, transition rules
(sync points, crossfade vs. stinger), and layers (vertical) or segments
(horizontal) explicitly. Untransitioned music cuts read as bugs.
Profile audio CPU and memory like any subsystem. Banks blow out memory
quietly; streaming vs. in-memory and compression codec choice (e.g. Vorbis vs.
ADPCM — CPU vs. size trade) are budget decisions, recorded as such.
Mix to a loudness target. Pick an integrated-loudness target (consoles
commonly target around −24 LUFS) and verify with a meter on real gameplay
capture, not isolated assets.
Audio architecture checklist
Middleware chosen and the decision recorded (Wwise / FMOD / engine-native), with the licensing-cost tier checked against budget
Bus tree drawn: master → music / SFX / dialogue / UI at minimum, with sidechain ducking dialogue over the rest
Fetches up-to-date documentation from Context7 for libraries and frameworks like React, Next.js, Prisma. Use for setup questions, API references, and code examples.