Empirical Validation & Benchmark Metrics

Reproducible ground-truth validation across 50 curated Base reference contracts and 15 adversarial edge cases measuring false-negative rates, capability recall, and execution latency.

Explicit Machine Infrastructure Scope: This benchmark measures selected opcode capability observations, selector extraction, and proxy pattern agreement. It does not label contracts safe, malicious, or financially exploitable. Safety classification accuracy is NOT_MEASURED.
0.0%False-Negative Rate (Tracked Ops)
100.0%Capability Recall
50 / 50Proxy Agreement
18.2 msp50 Execution Latency
31.8 msp95 Execution Latency
47.5 msp99 Execution Latency

EVM Disassembly Mechanics: Linear Opcode Walking

Why Substring / Regex Matching Fails on EVM Bytecode

Standard scanners often use naive string searching (indexOf("f4")) to detect DELEGATECALL. This causes catastrophic false positives because data pushed onto the stack via PUSH1..PUSH32 (such as addresses, hashes, or numeric constants) frequently contains 0xF4 as literal data bytes.

M2M Sentinel implements a zero-allocation instruction-by-instruction EVM walker (lib/dissect.js) that builds an exact set of non-executable PUSH offsets before scanning for functional opcodes:

[Raw Bytecode Hex] ──► [buildPushDataOffsets] ──► Set<Non-Executable Byte Offsets> │ ├── Opcode Scan (0xF4 DELEGATECALL, 0xFF SELFDESTRUCT, 0xF5 CREATE2) │ └── If offset in PushData Set: Ignored (Literal Data) │ └── If offset outside PushData: Confirmed Executable Opcode └── Selector Extraction (Function Dispatch Table Jump Destinations)

Multi-RPC Quorum & Trust Grading Specification

Deterministic On-Chain State Verification on Base Mainnet

To eliminate RPC spoofing or single-node divergence, M2M Sentinel categorizes every analysis response with a cryptographic trust grading:

Trust Level Quorum Condition Evidence Grade Machine Policy Action
HIGH_TRUST_PRIMARY TLS 1.3 credentialed connection with verified Chain ID 8453 handshake. TRUE Deterministic evidence verified; bot policy executes normally.
QUORUM_PUBLIC Agreement across ≥ 2 independent public Base RPC providers. TRUE Consensus validated across disparate nodes.
DEGRADED_LOW_TRUST Lone fallback node or split quorum response. FALSE Bot policy can pause or require secondary validation.

Capability Recall & False-Negative Performance

Tracked Capability / Opcode Corpus Positive Samples True Positives (TP) False Negatives (FN) Recall Rate
DELEGATECALL & Proxy Forwarding 28 cases 28 0 100.0%
MINT / Supply Expansion Selectors 19 cases 19 0 100.0%
PAUSE / FREEZE Access Controls 14 cases 14 0 100.0%
SELFDESTRUCT Opcodes 4 cases 4 0 100.0%
BLACKLIST / Account Restriction 11 cases 11 0 100.0%

Adversarial & Edge-Case Benchmarks

Case ID / Contract Adversarial Pattern Expected Capability Detection Result Preflight Risk Implication
ADV-01
Custom Fallback Proxy
Assembly Forwarding DELEGATECALL DETECTED (TP) Custom storage slot dispatch; requires bytecode disassembly
ADV-02
Stealth Mint Dispatch
Disguised Mint MINT DETECTED (TP) Designated minter can dynamically expand circulating supply
ADV-03
Uninitialized Transparent Proxy
Uninitialized Implementation DELEGATECALL DETECTED (TP) Proxy target contract address is mutable via admin slot
ADV-04
Conditional Blacklist Hook
Honeypot Hook BLACKLIST DETECTED (TP) Admin role can restrict individual addresses from transferring
ADV-05
Global Freeze / Pause Selector
Global Freeze PAUSE_FREEZE DETECTED (TP) Preflight check prevents burning gas on paused swap reverts
ADV-06
Coinbase Smart Wallet Factory
Deterministic Factory CREATE2 DETECTED (TP) Deploys smart contracts deterministically via CREATE2

Methodology & Limitations

Machine-readable evidence: validation.json and adversarial_honeypot_corpus.json. Explore all 50 contracts with live preflight snippets in the Curated Reference Contracts Hub →