[GH-ISSUE #3720] [Tracker] Emerging reverse-engineering domains: agentic AI, accelerators, confidential computing, SDV, IoT and next-generation runtimes #1430

New issue

Closed

opened 2026-10-16 21:05:22 +00:00 by bsakaguchi · 0 comments

bsakaguchi commented

2026-10-16 21:05:22 +00:00

Owner

Originally created by @matbrgz on GitHub (Oct 14, 2026).
Original GitHub issue: https://github.com/morluto/rea/issues/3720

Scope & Strategic Vision

Software systems developed for the 2026–2030 horizon are undergoing a fundamental architectural paradigm shift. Reverse engineering is expanding beyond traditional single-binary programs (binary -> CPU -> OS) into complex, composite systems composed of:

Agentic Multi-Agent Systems: Agent-to-Agent (A2A) protocols, streaming multi-turn dialogues, tool-use loops, and memory architectures.
On-Device AI & Specialized Accelerators: Edge neural processing units (NPU), ExecuTorch, Core AI Foundation Models, LiteRT, and host-device execution graphs.
Hardware-Enforced Confidential Computing: Attestation reports, TEE measurements (Intel TDX, AMD SEV-SNP, Arm CCA), and cryptographic workload bindings.
Post-Quantum Cryptography (PQC): NIST FIPS 203/204/205 algorithms (ML-KEM, ML-DSA, SLH-DSA) and hybrid migration patterns across binaries and protocols.
Software-Defined Vehicles (SDV) & Industrial IoT: AUTOSAR Adaptive architectures, SOME/IP, DDS, Matter/Thread smart-home mesh fabrics.
Modern Accelerated Web Runtimes: WebGPU shader pipelines (WGSL) and async WebAssembly Component Models (WASI 0.3).

This tracker coordinates REA's strategic capabilities across these forward-looking, next-generation reverse engineering domains.

Architectural Breakdown

flowchart TD
    A["Emerging Reverse Engineering Architecture (2026-2030)"] --> B["Tier 1: Agentic AI Protocols & Memory"]
    A --> C["Tier 2: Edge Accelerators & On-Device AI"]
    A --> D["Tier 3: Trust & Cryptographic Security (TEE & PQC)"]
    A --> E["Tier 4: Distributed & Physical Systems (SDV & IoT)"]
    A --> F["Tier 5: Modern Web & GPU Shaders (WebGPU & WASI 0.3)"]

    B --> B1["A2A Multi-Agent Workflows (#3721)"]
    B --> B2["LLM Applications, Tool Schemas & RAG (#3651)"]
    B --> B3["Tokenizers & Vocabulary Deconstruction (#3653)"]

    C --> C1["On-Device AI: Core AI & ExecuTorch (#3722)"]
    C --> C2["ML Model Topology: ONNX, TFLite, CoreML, TensorRT (#3637)"]
    C --> C3["ML Compiler IR: MLIR, StableHLO, XLA (#3673)"]
    C --> C4["PyTorch Checkpoints & Opcode Graphs (#3652)"]

    D --> D1["Confidential Computing & Attestation (#3723)"]
    D --> D2["Post-Quantum Cryptography Inventory (#3724)"]

    E --> E1["AUTOSAR Adaptive & SOME/IP Graphs (#3725)"]
    E --> E2["Matter & Thread Mesh Forensics (#3726)"]

    F --> F1["WebGPU Application & WGSL Graphs (#3727)"]
    F --> F2["Shader Bytecode: SPIR-V, DXBC, DXIL, Metal AIR (#3620)"]
    F --> F3["Wasm Components & Canonical ABI (#2350)"]

Complete Tracked Issues Index

1. Agentic AI Protocols & LLM Applications

#3721 — [Feature] A2A protocol reverse engineering: Agent Cards, task lifecycles, artifacts, streaming sessions and multi-agent workflow reconstruction
#3651 — [Feature] LLM application reverse engineering: system prompts, model/provider routing, tool schemas, RAG links and agent workflow reconstruction
#3653 — [Feature] Tokenizer and vocabulary reverse engineering: BPE merges, WordPiece models, SPM unigrams, special tokens and regex splitters

2. Edge Accelerators, ML Runtimes & Model Topologies

#3722 — [Feature] On-device AI runtime reverse engineering: Core AI, Foundation Models, ExecuTorch, LiteRT and NPU delegate reconstruction
#3637 — [Feature] ML model graph reverse engineering: ONNX, TFLite, CoreML, and TensorRT topology
#3652 — [Feature] PyTorch checkpoint and model artifact reverse engineering: ZIP/pickle opcode graph, tensor storages, state_dict reconstruction and unsafe-global detection
#3673 — [Feature] ML compiler IR reverse engineering: MLIR bytecode, StableHLO, TOSA and XLA HLO graph reconstruction

3. Hardware Trust & Cryptography

#3723 — [Feature] Confidential-computing reverse engineering: TDX, SEV-SNP, Arm CCA attestation, measurements and workload binding
#3724 — [Feature] Post-quantum cryptography reverse engineering: ML-KEM, ML-DSA and SLH-DSA fingerprints, parameters and migration inventory

4. Software-Defined Vehicles (SDV) & Industrial IoT

#3725 — [Feature] Software-defined vehicle reverse engineering: AUTOSAR Adaptive, SOME/IP, DDS, DoIP, UDS and service graphs
#3726 — [Feature] Matter & Thread reverse engineering: commissioning captures, TLV, clusters, endpoints, fabrics and mesh topology reconstruction

5. WebGPU, Shader Bytecode & Modern Web Runtimes

#3727 — [Feature] WebGPU application reverse engineering: WGSL, pipelines, bind groups, compute passes and JavaScript-to-GPU graphs
#3620 — [Feature] Shader and GPU Bytecode Reverse Engineering: SPIR-V, DXBC, DXIL, and Metal AIR reflection and decompilation
#2350 — [Feature] Inspect WebAssembly components and canonical ABI declarations offline

Originally created by @matbrgz on GitHub (Oct 14, 2026). Original GitHub issue: https://github.com/morluto/rea/issues/3720 [![Type: Tracker](https://img.shields.io/badge/type-tracker-purple.svg?style=flat-square)](https://github.com/morluto/rea) [![Status: Proposed](https://img.shields.io/badge/status-proposed-blue.svg?style=flat-square)](https://github.com/morluto/rea) [![Ecosystem: REA](https://img.shields.io/badge/ecosystem-morluto%2Frea-8A2BE2.svg?style=flat-square)](https://github.com/morluto/rea) ### Scope & Strategic Vision Software systems developed for the 2026–2030 horizon are undergoing a fundamental architectural paradigm shift. Reverse engineering is expanding beyond traditional single-binary programs (`binary -> CPU -> OS`) into complex, composite systems composed of: - **Agentic Multi-Agent Systems**: Agent-to-Agent (A2A) protocols, streaming multi-turn dialogues, tool-use loops, and memory architectures. - **On-Device AI & Specialized Accelerators**: Edge neural processing units (NPU), ExecuTorch, Core AI Foundation Models, LiteRT, and host-device execution graphs. - **Hardware-Enforced Confidential Computing**: Attestation reports, TEE measurements (Intel TDX, AMD SEV-SNP, Arm CCA), and cryptographic workload bindings. - **Post-Quantum Cryptography (PQC)**: NIST FIPS 203/204/205 algorithms (ML-KEM, ML-DSA, SLH-DSA) and hybrid migration patterns across binaries and protocols. - **Software-Defined Vehicles (SDV) & Industrial IoT**: AUTOSAR Adaptive architectures, SOME/IP, DDS, Matter/Thread smart-home mesh fabrics. - **Modern Accelerated Web Runtimes**: WebGPU shader pipelines (WGSL) and async WebAssembly Component Models (WASI 0.3). This tracker coordinates REA's strategic capabilities across these forward-looking, next-generation reverse engineering domains. ### Architectural Breakdown ```mermaid flowchart TD A["Emerging Reverse Engineering Architecture (2026-2030)"] --> B["Tier 1: Agentic AI Protocols & Memory"] A --> C["Tier 2: Edge Accelerators & On-Device AI"] A --> D["Tier 3: Trust & Cryptographic Security (TEE & PQC)"] A --> E["Tier 4: Distributed & Physical Systems (SDV & IoT)"] A --> F["Tier 5: Modern Web & GPU Shaders (WebGPU & WASI 0.3)"] B --> B1["A2A Multi-Agent Workflows (#3721)"] B --> B2["LLM Applications, Tool Schemas & RAG (#3651)"] B --> B3["Tokenizers & Vocabulary Deconstruction (#3653)"] C --> C1["On-Device AI: Core AI & ExecuTorch (#3722)"] C --> C2["ML Model Topology: ONNX, TFLite, CoreML, TensorRT (#3637)"] C --> C3["ML Compiler IR: MLIR, StableHLO, XLA (#3673)"] C --> C4["PyTorch Checkpoints & Opcode Graphs (#3652)"] D --> D1["Confidential Computing & Attestation (#3723)"] D --> D2["Post-Quantum Cryptography Inventory (#3724)"] E --> E1["AUTOSAR Adaptive & SOME/IP Graphs (#3725)"] E --> E2["Matter & Thread Mesh Forensics (#3726)"] F --> F1["WebGPU Application & WGSL Graphs (#3727)"] F --> F2["Shader Bytecode: SPIR-V, DXBC, DXIL, Metal AIR (#3620)"] F --> F3["Wasm Components & Canonical ABI (#2350)"] ``` --- ## Complete Tracked Issues Index ### 1. Agentic AI Protocols & LLM Applications - [ ] #3721 — `[Feature] A2A protocol reverse engineering: Agent Cards, task lifecycles, artifacts, streaming sessions and multi-agent workflow reconstruction` - [ ] #3651 — `[Feature] LLM application reverse engineering: system prompts, model/provider routing, tool schemas, RAG links and agent workflow reconstruction` - [ ] #3653 — `[Feature] Tokenizer and vocabulary reverse engineering: BPE merges, WordPiece models, SPM unigrams, special tokens and regex splitters` ### 2. Edge Accelerators, ML Runtimes & Model Topologies - [ ] #3722 — `[Feature] On-device AI runtime reverse engineering: Core AI, Foundation Models, ExecuTorch, LiteRT and NPU delegate reconstruction` - [ ] #3637 — `[Feature] ML model graph reverse engineering: ONNX, TFLite, CoreML, and TensorRT topology` - [ ] #3652 — `[Feature] PyTorch checkpoint and model artifact reverse engineering: ZIP/pickle opcode graph, tensor storages, state_dict reconstruction and unsafe-global detection` - [ ] #3673 — `[Feature] ML compiler IR reverse engineering: MLIR bytecode, StableHLO, TOSA and XLA HLO graph reconstruction` ### 3. Hardware Trust & Cryptography - [ ] #3723 — `[Feature] Confidential-computing reverse engineering: TDX, SEV-SNP, Arm CCA attestation, measurements and workload binding` - [ ] #3724 — `[Feature] Post-quantum cryptography reverse engineering: ML-KEM, ML-DSA and SLH-DSA fingerprints, parameters and migration inventory` ### 4. Software-Defined Vehicles (SDV) & Industrial IoT - [ ] #3725 — `[Feature] Software-defined vehicle reverse engineering: AUTOSAR Adaptive, SOME/IP, DDS, DoIP, UDS and service graphs` - [ ] #3726 — `[Feature] Matter & Thread reverse engineering: commissioning captures, TLV, clusters, endpoints, fabrics and mesh topology reconstruction` ### 5. WebGPU, Shader Bytecode & Modern Web Runtimes - [ ] #3727 — `[Feature] WebGPU application reverse engineering: WGSL, pipelines, bind groups, compute passes and JavaScript-to-GPU graphs` - [ ] #3620 — `[Feature] Shader and GPU Bytecode Reverse Engineering: SPIR-V, DXBC, DXIL, and Metal AIR reflection and decompilation` - [ ] #2350 — `[Feature] Inspect WebAssembly components and canonical ABI declarations offline`

bsakaguchi closed this issue

2026-10-16 21:05:22 +00:00

bsakaguchi referenced this issue

2026-10-17 21:11:22 +00:00

[PR #1430] [MERGED] perf(native): record macOS command provenance once in its Evidence #2882

bsakaguchi referenced this issue

2026-10-17 21:11:25 +00:00

[PR #1436] [MERGED] test(native): verify macOS provenance through real CLI and MCP #2885