Drew T
f5f4c2eeec
Two banks from the S75 redraft workflow (7 overlay functions, one agent each,
every claimed MATCH re-verified by an independent agent instructed to refute
it). Both were carried as F-FAR "a draft exists but is materially wrong":
func_801806F8 ov_SC03_105 241 ins (recorded closeness 235)
func_80180ABC ov_SC03_105 257 ins (recorded closeness 250)
Neither needed a better model. Both needed the recorded closeness not to be
believed -- see below.
frontier_classify.py, THREE fixes, each caught by testing against a case whose
answer was already known:
1. BEST closeness, not LAST. .run/backlog.jsonl is append-only, one row per
attempt across every lane and session, so the last row is evidence about
THAT LANE'S SEED, not about the function. Caught func_80180B3C (best 125,
last 287) and moved func_80181294 from "redraft" to "permuter" (best 19).
The draft that ACHIEVED the best score is kept, not the last one written.
2. journal_notes.py wired in as a SECOND, DISAGREEING oracle (R34). The backlog
does not have what the agent journals have. Measured on func_8017DB98:
backlog best == last == 115, so best-vs-last could not help, while the
journal holds "Attempt 2 (MATCH · closeness 0) ... MATCH 122/122 ... BANK
BLOCKER is TU plumbing, not the body (§376/§378)" WITH the draft path and the
exact declaration to change. Reclassified 37 functions; G-DRAFTED-UNKNOWN
fell 47 -> 10 and a new C-PLUMBING class holds 16 functions / 1,547 ins whose
BODIES ARE PROVEN and are blocked only by the TU.
3. A consuming-regex bug in my own extractor -- the session's signature defect,
committed a third time in the tool written to find it. The first cut used
`re.finditer(r'\*\*Attempt \d+\*\* \(([^)]*)\)(.{0,400})', ..., re.S)`, whose
400-char body window SWALLOWS THE NEXT ATTEMPT'S HEADER, so every record
following another was invisible. On func_8017DB98 it hid attempts 2 AND 6,
both `MATCH · closeness 0`, and returned attempt 1's NEAR (2) as the best --
exactly the records the oracle exists to find. Now splits on the marker
rather than consuming past it. A regex that consumes an unbounded body cannot
enumerate the items after the first.
Rows now carry attempts, closeness_last, journal_closeness, and a
!!WARMSTART-REGRESSION flag when a later attempt scored materially worse than
the best -- the shape a wave's warm-start regression makes, which from inside
the wave is indistinguishable from an unsolved function.
Gate ledger for the batch of 7: 2 banked, 3 near, 2 failed. func_800CB00C failed
despite being adversarially upheld -- it owns a jump table, and both matchers
compare .text only, so a verified .text MATCH proves nothing about table
placement (the agent's own write-up says so).
BFM-decomp
A matching decompilation of Brave Fencer Musashi (PlayStation, SLUS-00726, USA 1998) — the first public decompilation effort for this game.
What "matching" means
The goal is C source code that, compiled with the original-era toolchain (PsyQ 4.x / GCC 2.7.2-family + ASPSX via maspsx), produces a byte-for-byte identical SLUS_007.26 and, eventually, byte-identical overlay binaries. SHA1 checksums are the ground truth; "functionally equivalent" does not count.
No ROM content
This repository contains no game assets, no disassembly output, and no ROM-derived data — only source code, build configuration, symbol names/addresses, hashes, and documentation. To build or contribute you must provide your own dump of the game disc (4-track BIN/CUE, redump layout). See .gitignore for the firewall.
Project status
Latest (Phase 19, 2026-06-20): the project builds 136 binaries byte-identical from a clean tree (the EXE + the resident engine + all 134 location overlays);
make check-all→ 136/136. Fleet byte-identical-from-source is 58.0% (function-instance-weighted; see the PhaseEnds for the byte-weighted ~30% figure and what it includes). Shared engine functions are matched once inov_SC01_077and propagated ×134 viatools/dedup_propagate.py. (The narrative below is Phase-11/12-era; a full refresh is part of the public-flip prep.)
Gen1 (foundation) complete — the matching pipeline is proven end-to-end. make extract && make build && make check rebuilds SLUS_007.26 byte-for-byte identical (SHA1 143dbb89…) from C + assembly, reproducibly across many sessions.
- Compiler pinned by evidence:
gcc-2.7.2-psx -O2 -G0 -mips1 -mcpu=3000+maspsx --aspsx-version=2.56 --expand-div. - 52 functions hand-matched to byte-identical machine code — including the LZSS streaming decompressor — with a decomp-permuter + matching-cookbook "flywheel" to accelerate the next.
- 959 PsyQ SDK functions linked byte-identical (libcd, libgs, libgte, libspu/libsnd, libgpu, libc2, libmcrd, libapi/libcard, libetc) straight from the real PsyQ 4.0 libraries instead of re-decompiling them — bringing byte-identical-from-source coverage of the EXE to ~50%.
- File-loader / overlay system reverse-engineered, with the resident engine blob + location overlays' load addresses proven byte-identical against a live PCSX-Redux RAM dump.
About half the EXE is still INCLUDE_ASM stubs (correct bytes, not yet C), and the bulk of the game lives in compressed overlays inside the .CD archives — Gen2 (overlays & engine at scale) is underway:
- The build toolchain is binary-agnostic (one parameterized pipeline builds any binary), and the always-resident engine blob rebuilds byte-for-byte from source (SHA1
8e17e02f…) — the second binary reconstructed exactly, after the EXE — and is now 86% hand-matched C (123 / 146 functions, up from 0): its scripting turned out to be compiled-MIPS state/mode dispatch, not a bytecode VM, and the save-file + sound (SQV) formats are documented. The harvest used a reusable swarm-of-agents + bit-for-bit byte-gate method (a wrong match can't be accepted) —tools/harvest_verify.py+tools/match_one.py, which carry straight into the overlay phase. - A cross-binary deduplication pipeline is live: a Ghidra-free signer fingerprints all 134 location overlays, and the report finds ~9,000 byte-identical function groups shared across binaries (~28 MB of collapsible code) — a single engine function is byte-identical in all 134 overlays. This is "one match unlocks many": each engine match will be auto-credited across the overlay fleet.
Current phase and detailed progress live in phase-ends/ (newest PhaseEnd_*.md = current state); methodology, rules, and the full roadmap are in PROJECT_CONTEXT.md; environment setup in docs/SETUP.md.
This project is developed primarily by Claude Code driving Ghidra through an MCP server; see CLAUDE.md.
License
Private repository for now. AGPL-3.0 is planned at public release, modeled on sotn-decomp. tools/brave-CUE/ is CUE's BRAVE extractor (GPL, source included) and retains its own license.