Brisbane, AustraliaIndependent research & engineering

Researching how software is analysed, transformed and protected.

AUSTRAL develops low-level program-analysis infrastructure and applies it across native transformation, adversarial evaluation and practical software-protection programmes.

Research thesis

Protection is a program-analysis problem first.

Durable protection depends on precise representations, controlled transformations, reliable lowering and evidence that the result still behaves correctly.

01

Understand before protecting

Protection begins with accurate models of execution, state, control flow and platform behaviour. Those models guide every transformation.

02

Correctness is a release requirement

Every pass must preserve intended behaviour across real toolchains, build configurations and target environments.

03

Attack the work ourselves

Deobfuscation and reverse-engineering workflows are built into the evaluation loop from the first implementation.

Current programmes

A focused research stack.

Each programme shares infrastructure and validation work. New domains follow demonstrated capability in the underlying stack.

R / 01Foundation · active

Native transformation framework

A reusable x86-64 analysis and rewriting foundation: decoding, control-flow recovery, symbolic representation, transformation passes, lowering and validation.

x86-64symbolic IRCFG analysisbinary rewriting
R / 02Applied · active

Godot ecosystem protection

Applied protection engineering for Godot projects, covering GDScript, resources and extension compatibility on top of the wider analysis stack.

GDScriptresourcesGDExtensioncompatibility
R / 03Research track

Deobfuscation and evaluation

Analysis tooling used to challenge protection strategies, simplify transformed programs and measure what creates durable cost for reverse engineers.

liftingsimplificationdynamic analysisevaluation
Explore Research
Engineering method

Human-directed. AI-accelerated. Evidence-bound.

AUSTRAL uses specialised AI systems to increase research throughput. Architecture, methodology, interpretation and release decisions remain human responsibilities.

01

Specify

Define invariants, threat models, compatibility constraints and acceptable failure conditions before implementation.

02

Build and review

Use focused agents for implementation, tests, documentation and fresh-context review against authorised technical sources.

03

Validate

Run reproducible experiments, differential tests and adversarial analysis. Claims are limited to what the evidence supports.

Research before product

Commercial tooling is downstream of validated infrastructure and measured research.

Compatibility by design

Protection earns its place through stable output, owner visibility and long-term supportability.

Evidence-led publication

Publication begins when the method, evidence and technical record are ready for scrutiny.

Technical records

Publish what survives scrutiny.

AUSTRAL will publish research notes, engineering records and programme updates as the work becomes reproducible. Each release pairs its claims with clear methodology and evidence.

Publication standard

AUSTRAL / RECORDS
01A clearly scoped research question
02Reproducible methodology and artefacts
03Measured results with stated limitations
04Human review of every published claim

The first technical records are in preparation across the native analysis, transformation and evaluation programmes.

Australian capability

Built in Australia for a global software landscape—independent, technically ambitious and deliberately long-term.

AUSTRAL is being developed as an independent research organisation spanning software protection, program analysis and future game-integrity research.