A DECOMPILER IS ONLY STAGE ONE

The eight-stage EX5 evidence pipeline

Most tools stop at parsing a file or emitting raw disassembly. Our backend continues through eight coordinated stages, each forced to agree with the last — turning an opaque binary into a confidence-scored technical case.

FingerprintMapObserveLiftExploreReasonChallengeScoreCase
“Most tools give you a view. This pipeline is built to build a technical case.”
01
FINGERPRINT
Characterize the artifact

Identify container properties, architecture signals, integrity conditions, compiler and build indicators, and version-sensitive constraints — the profile everything downstream is calibrated against.

reads .ex5 byteswrites build profileengines Custom .NET parsers
02
MAP
Recover the structure

Recover accessible code and data regions, candidate functions, references, call relationships, control flow, constants, strings, and platform interactions into a navigable structural map.

reads fingerprint + byteswrites functions · CFG · xrefsengines IDA · Ghidra · radare2
03
OBSERVE
Confirm behavior at runtime

Where the target and authorization permit, scripted instrumentation collects reproducible execution evidence — transitions, register and memory state, module and call-path activity — under defined, logged conditions.

reads structural mapwrites runtime tracesengines x64dbg · WinDbg · Frida · DynamoRIO · Pin
04
LIFT
Normalize into IR

Lift low-level instructions into an architecture-aware intermediate representation modeling blocks, calls, state, types, and side effects — the common language every later stage reasons over.

reads static + runtime evidencewrites IR + evidence graphengines Proprietary IR
05
EXPLORE
Prune the candidate space

Evaluate competing interpretations under structural, data-flow, type, calling-convention, platform, and runtime constraints — eliminating everything the evidence forbids and ranking what survives.

reads IR + constraintswrites ranked candidatesengines Constraint solver
06
REASON
Apply specialist AI

Authorized high-capability models classify routines, infer types and roles, correlate distant evidence, and flag ambiguity — proposing hypotheses, never declaring verdicts.

reads ranked candidateswrites semantic hypothesesengines OpenAI · Anthropic APIs
07
CHALLENGE
Force the evidence to agree

Compare independent passes, expose contradictions, reject incompatible hypotheses, and return ambiguous regions for deeper analysis before anything is allowed to progress.

reads hypotheses + evidencewrites reconciled findingsengines Differential validators
08
SCORE
Deliver with confidence

Label each finding by evidentiary strength — verified, high, probable, tentative, or unresolved — so a strong result in one region never disguises uncertainty in another.

reads reconciled findingswrites confidence-scored .mq5engines Scoring + report gate
WHY EIGHT STAGES

No single tool decides the answer

One disassembler gives a view. One debugger gives a trace. One model gives an opinion. The pipeline correlates them, rejects candidates the binary forbids, records provenance, and only promotes a finding once independent evidence agrees.

Evidence from every stage is correlated in one shared representation.
Candidates that violate binary constraints are rejected, not smoothed over.
Provenance and blockers are recorded at each stage boundary.
Confidence is assigned per finding — conflicts lower it instead of hiding.

Run all eight stages on your own EX5

Review the published license terms and preview the planned checkout experience for the local Windows workflow. The current demo does not accept payment or issue software.

Review licensingBack to overview