8-Target Benchmark Suite

Unprotected Baseline vs. BLA Protected Output

Run all 8 benchmark targets locally with npx bla-agents@latest benchmark.

Target 01 · Electron .asar

Notion Desktop Clipboard Bridge

Baseline: Analyzed in 38s ($0.20).
Protected: IPC de-literalized + 22 obligation families + shadow logic → 41.3x token ratio ($14.88).

View Verification Output

• ipcChannelsPairedUniquely: 0 • triggeredLimits: ["alias-selections", "primitive-candidates", "primitive-bytes", "getter-expansion"] • runtimeEquivalence: 100% Exact

Target 02 · JavaScript Audio/Physics

DX-Ball Stereo Sound-Pan Engine

Baseline: Analyzed in 22s ($0.13).
Protected: 9-chunk opaque ternaries + 40-node SCC mesh → 49.2x token ratio ($11.45).

View Verification Output

• primitiveCandidateBranches: 512 (> 256 limit) • sccLabyrinthEdges: 160 • runtimeEquivalence: calculateSoundPan(160, 640) === -0.5

Target 03 · Native ELF64 / Mach-O / PE

Enterprise License & Fuse Daemon

Baseline: Crates & fuses extracted in 45s.
Protected: 72 rustc hashes (>64 limit) + 33 fuse sentinels (>32 limit) → Parser Limit Reached.

View Verification Output

• RustBinaryProvider: exceeds MAX_RUSTC_HASHES (64) • ElectronFuseReader: exceeds maximumSentinelsPerSlice (32) • Native exports preserved: 100%

Target 04 · SQLite & AI Models

.sqlite, .safetensors & .gguf Vaults

Baseline: Schema & tensor headers parsed in 19s.
Protected: Virtual generated-column guard + 35-level metadata nesting → 38.6x token ratio.

View Verification Output

• SQLite: SqliteColumnEvaluationFailure on read-only connection • Safetensors/GGUF: ModelFormatFailure("capacity", >32 levels) • Existing GGUF KV & tensors preserved: 100%