Back to skill
v1.0.0

quantum-sim

BenignClawScan verdict for this skill. Analyzed May 1, 2026, 6:49 AM.

Analysis

This appears to be a coherent local quantum circuit simulator with no artifact-backed credential, network, or persistence behavior, but it runs a bundled Python script and large qubit counts may strain system resources.

GuidanceThis skill looks suitable for local quantum-circuit experiments. Before installing, be comfortable with running the bundled Python script, and avoid very large qubit counts unless you are prepared for high CPU or memory use.

Findings (2)

Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.

Abnormal behavior control

Checks for instructions or behavior that redirect the agent, misuse tools, execute unexpected code, cascade across systems, exploit user trust, or continue outside the intended task.

Unexpected Code Execution
SeverityInfoConfidenceHighStatusNote
SKILL.md
python3 {baseDir}/scripts/quantum_sim.py --preset bell

The skill is intended to work by running a bundled Python simulator. This local code execution is clearly disclosed and directly aligned with the quantum-simulation purpose.

User impactInvoking the skill may run the included Python script on the user's machine to perform simulations.
RecommendationUse it when local script execution is acceptable; review or restrict execution if you have strict local-code policies.
Tool Misuse and Exploitation
SeverityLowConfidenceMediumStatusNote
scripts/quantum_sim.py
for i in range(n-1, -1, -1):
                if i == qubit:
                    op = np.kron(op, np.array(gate_mat, dtype=complex))
...
            self.state = op @ self.state

The visible single-qubit gate implementation builds a full operator matrix before applying it. For high qubit counts, this can use far more CPU and memory than users may expect from a simple statevector simulator.

User impactLarge simulations, especially near the advertised 20-qubit limit, may hang or exhaust local memory.
RecommendationStart with small qubit counts and avoid unattended large runs; the skill author should document practical limits or use an in-place statevector update implementation.