rb3

Orchestrate a multi-wave, doc-synced investigation loop into (1) AUDIO CORRECTNESS — record the game's output and measure how far it diverges from the expected/ground-truth source audio (the "clipped noise" problem), and (2) ASSET-LOAD PERFORMANCE — profile and attribute the load-time stutters. You act as orchestrator/coordinator: each wave plans+reviews, fans out subagents (via the Workflow tool / ultracode) over independent hypotheses, converges, builds, re-measures, then plans the next wave. Agents hand off through on-disk docs. Use when iterating on web/native audio fidelity or load-stutter perf.

Verify the game's recorded audio is the SAME audio as the source song AND is played CORRECTLY (right pitch/speed, not clipped/noise/silent). Decrypts+decodes the source mogg into a ground-truth reference, captures the game output headless, then measures identity (mix-robust chroma cross-correlation + Chromaprint fingerprint), playback rate (resample-search → speed ratio; catches the 'chipmunk' bug), and distortion (reference-free clip/flat-top/wrap). Use after any audio change (mixer, limiter, resampler, decode, gain) to confirm the right song still plays correctly, or to investigate an audio-fidelity report. Native-first; works for web captures too.

Diff stack-frame layouts between target and base for a function. Labels base-side slots with source variable names from a MWCC DWARF recompile. Identifies SWAPPED pairs (decl-reorder candidates), SHIFTED slots, DIFFER (different variables in same slot), and TGT_ONLY / BASE_ONLY (extra/missing locals). Filters out callee-save slots.

Diff a DATA symbol (vtable, pointer/jump table, string pool, static initializer) between the target and the decompiled build, showing byte differences and — for each relocation slot — which function/symbol each side points to. Use when a data symbol is below 100%, or to see exactly where your decompiled vtable diverges from the target (which slot resolves to the wrong function). This is the diff counterpart to /vtable and /resolve-vcall, which only read the target.

Resolve a single virtual-call slot to its function symbol. Given `class, sub-object offset, slot`, returns the mangled symbol at that slot of the named class's MWCC vtable. Use when objdiff shows a vcall mismatch and you have the three numbers off the asm but don't want to dump the whole vtable.

Dump a class's MWCC vtable from build/SZBE69_B8/obj/*.o. Shows every sub-object table (primary + each base), slot index, byte offset, and the mangled function symbol. Use when objdiff shows a `lwz r12, OFF(r12)` mismatch and you need to map the offset to a virtual method, or when verifying a header's virtual-function order against the compiled binary.

Get decomp progress summary. Shows match statistics from the build report, per-category breakdowns, and identifies areas with the most remaining work. Use to check overall project status or find targets for decomp work.

Sweep a unit or glob of units for already-matching functions. Reads build/SZBE69_B8/report.json (no per-function objdiff invocations), auto-marks 100% matches as COMPLETE in decomp.db, and reports partial-match candidates ranked by closeness to 100%. Use this after a header change, a build-system change, or to find a unit's untouched workable functions.

Full analysis of a function combining objdiff match%, Ghidra decompilation, and m2c output. The go-to command before starting decomp work. Shows current state, target code structure, and a starting point for implementation.

Side-by-side target vs base assembly comparison with mismatch annotations. Shows aligned instructions with match markers, register swap details, and offset deltas. Use when diagnosing the last few percent of a function.

Find DC3 reference implementation for an RB3 unit. Shows function overlap, DC3 match status, and optionally the DC3 source code. Use to leverage DC3's ~87% complete decomp for shared Milo engine code.

Decompile a function from Ghidra (RB3 Wii debug ELF). Shows decompiled C code with full DWARF symbols. Use when investigating function structure, types, or control flow before starting decomp work.

Clean up decomp code after a first pass. Improve readability and maintainability while preserving exact match percentage. Used as a second pass to polish rough implementations.