Skip to content
Back to skills

Move Smith Fuzzer

ASecurity

Move Smith Fuzzer Skill

  • 61 stars
  • 0 votes
  • 0 copies
  • 3 views
  • Added May 29, 2026
toolspythongobashtesting

Works with

  • mcp

Security analysis

A100/100

Pro scans all 2 files and shows the line behind each finding

Scanned May 29, 2026

npx -y skills add plurigrid/asi --skill move-smith-fuzzer --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Move Smith Fuzzer?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Move Smith Fuzzer
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/plurigrid-move-smith-fuzzer/badge)](https://www.skillsdirectory.com/skills/plurigrid-move-smith-fuzzer)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: move-smith-fuzzer
description: Move Smith Fuzzer Skill
version: 1.0.0
---


# move-smith-fuzzer Skill


> *"Find bugs before they find your users. Fuzzing as validation."*

## Overview

**Move Smith Fuzzer** implements property-based testing and fuzzing for Move smart contracts. Uses MoveSmith's differential testing against multiple Move VMs to find consensus-breaking bugs.

## GF(3) Role

| Aspect | Value |
|--------|-------|
| Trit | -1 (MINUS) |
| Role | VALIDATOR |
| Function | Validates Move contracts via fuzz testing |

## Architecture

```
┌─────────────────────────────────────────────────────────────────┐
│                    MOVE SMITH FUZZER                            │
├─────────────────────────────────────────────────────────────────┤
│                                                                 │
│  Contract Source    Generator      Fuzzer         Report       │
│  (+1 GEN)          (0 COORD)      (-1 VAL)        (output)     │
│      │                 │              │               │        │
│      ▼                 ▼              ▼               ▼        │
│  ┌───────┐        ┌────────┐    ┌──────────┐   ┌─────────┐    │
│  │ Parse │───────►│Generate│───►│ Execute  │──►│ Report  │    │
│  │ AST   │        │ Inputs │    │ & Compare│   │ Bugs    │    │
│  └───────┘        └────────┘    └──────────┘   └─────────┘    │
│                                      │                         │
│                                      ▼                         │
│                              ┌──────────────┐                  │
│                              │ Differential │                  │
│                              │   Testing    │                  │
│                              └──────────────┘                  │
│                                      │                         │
│                         ┌────────────┼────────────┐            │
│                         ▼            ▼            ▼            │
│                    Move VM 1    Move VM 2    Move VM 3        │
│                                                                │
└─────────────────────────────────────────────────────────────────┘
```

## MoveSmith Integration

```python
class MoveSmithFuzzer:
    """
    MoveSmith-based differential fuzzer for Move.

    Reference: "MoveSmith: Compiler Bug Isolation via Compilation
    Result Consistency Checking for Move"
    """

    TRIT = -1  # VALIDATOR role

    def __init__(self, vms: list[MoveVM]):
        self.vms = vms  # Multiple VMs for differential testing
        self.mutator = MoveMutator()
        self.oracle = DifferentialOracle()

    def fuzz(self, contract: str, iterations: int = 10000) -> list:
        """
        Fuzz a Move contract for bugs.

        Strategy:
        1. Generate random valid Move programs
        2. Execute on all VMs
        3. Compare results (differential testing)
        4. Report discrepancies
        """
        bugs = []

        for i in range(iterations):
            # Generate or mutate
            if random.random() < 0.3:
                program = self.generate_random_program()
            else:
                program = self.mutator.mutate(contract)

            # Execute on all VMs
            results = {}
            for vm in self.vms:
                try:
                    results[vm.name] = vm.execute(program)
                except Exception as e:
                    results[vm.name] = ('error', str(e))

            # Differential testing
            if not self.oracle.consistent(results):
                bugs.append({
                    'program': program,
                    'results': results,
                    'type': 'differential',
                    'iteration': i
                })

        return bugs


class MoveMutator:
    """Mutates Move programs to find edge cases."""

    def mutate(self, program: str) -> str:
        """Apply random mutations."""
        mutations = [
            self.mutate_integers,
            self.mutate_addresses,
            self.mutate_vectors,
            self.mutate_structs,
            self.swap_operations,
        ]

        mutated = program
        for _ in range(random.randint(1, 5)):
            mutation = random.choice(mutations)
            mutated = mutation(mutated)

        return mutated

    def mutate_integers(self, program: str) -> str:
        """Replace integers with edge cases."""
        edges = [0, 1, 255, 256, 2**64-1, 2**128-1]
        # Find and replace integer literals
        return re.sub(r'\b(\d+)\b',
                     lambda m: str(random.choice(edges)),
                     program)
```

## Prover-Fuzzer Synergy

```python
class ProverFuzzerHybrid:
    """
    Combine Move Prover with fuzzing.

    Prover: Proves properties hold for ALL inputs
    Fuzzer: Finds counterexamples for SOME inputs

    Together: Maximum coverage
    """

    def verify_contract(self, contract: str) -> dict:
        # First: Prover for formal guarantees
        prover_result = move_prover.verify(contract)

        # Second: Fuzzer for edge cases prover missed
        fuzzer_bugs = self.fuzzer.fuzz(contract)

        return {
            'proven_properties': prover_result.properties,
            'fuzzer_bugs': fuzzer_bugs,
            'confidence': self.compute_confidence(prover_result, fuzzer_bugs)
        }
```

## Property-Based Testing

```move
#[test]
fun test_gf3_conservation() {
    let seed = 0x42D;
    let mut prng = movemate_random::new(seed);

    for i in 0..1000 {
        // Generate random triads
        let trit1 = random_trit(&mut prng);
        let trit2 = random_trit(&mut prng);
        let trit3 = (3 - trit1 - trit2) % 3;  // Force conservation

        // Property: sum must be 0 mod 3
        let sum = (trit1 + trit2 + trit3) % 3;
        assert!(sum == 0, 0);
    }
}

fun random_trit(prng: &mut PRNG): u8 {
    movemate_random::next_u8(prng) % 3
}
```

## Differential Testing VMs

| VM | Purpose | Speed |
|----|---------|-------|
| Aptos Move VM | Production reference | Medium |
| Move VM (reference) | Original implementation | Slow |
| Revela decompiler | Bytecode analysis | Fast |
| MoveSmith interpreter | Fuzzing-optimized | Fast |

## GF(3) Triads

```
move-smith-fuzzer (-1) ⊗ move-narya-bridge (0) ⊗ aptos-gf3-society (+1) = 0 ✓
move-smith-fuzzer (-1) ⊗ datalog-fixpoint (0) ⊗ discopy (+1) = 0 ✓
move-smith-fuzzer (-1) ⊗ interaction-nets (0) ⊗ gay-mcp (+1) = 0 ✓
```

## Commands

```bash
# Fuzz a Move module
just move-fuzz sources/gf3.move --iterations 10000

# Differential testing across VMs
just move-diff sources/gf3.move --vms aptos,reference

# Property-based test with random seeds
just move-proptest sources/ --seed 0x42D

# Generate coverage report
just move-fuzz-coverage sources/ --output coverage.html
```

---

**Skill Name**: move-smith-fuzzer
**Type**: Fuzzing / Property-Based Testing
**Trit**: -1 (MINUS - VALIDATOR)
**GF(3)**: Validates Move contracts through differential testing


## Scientific Skill Interleaving

This skill connects to the K-Dense-AI/claude-scientific-skills ecosystem:

### Graph Theory
- **networkx** [○] via bicomodule
  - Universal graph hub

### Bibliography References

- `general`: 734 citations in bib.duckdb

## Cat# Integration

This skill maps to Cat# = Comod(P) as a bicomodule in the Prof home:

```
Trit: 0 (ERGODIC)
Home: Prof (profunctors/bimodules)
Poly Op: ⊗ (parallel composition)
Kan Role: Adj (adjunction bridge)
```

### GF(3) Naturality

The skill participates in triads where:
```
(-1) + (0) + (+1) ≡ 0 (mod 3)
```

This ensures compositional coherence in the Cat# equipment structure.

Files in this skill

  • README.md7.6 KB
  • SKILL.md8.3 KB

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.