AgentReadyHomeAgent Listing

← Bonsai Image

Bonsai Image — agentic threat model

6.7AIVSS 6.7 · Medium

Bonsai Image is a low-risk, single-purpose image generation tool with minimal agentic capabilities, meaning its primary security concerns are restricted to model-level vulnerabilities (e.g., adversarial prompts, unsafe content generation) and client-side execution risks.

OWASP AIVSS score rationale

AIVSS = (CVSS_Base + AARS) × Mitigation_Factor, where AARS = (10 − CVSS_Base) × (Factor_Sum / 10) × ThM
CVSS base 6.1AARS uplift 0.59Factor sum 1.5/10Threat ×1.0Mitigation ×1.0
Autonomy of Action
0.10
Goal-Driven Planning
0.00
Self-Modification
0.00
Dynamic Tool Use
0.00
Persistent Memory
0.00
Contextual Awareness
0.10
Dynamic Identity
0.00
Multi-Agent Interactions
0.00
Non-Determinism
0.60
Opacity & Reflexivity
0.70

Scored with the canonical OWASP AIVSS formula (AIVSS calculator reference); agentic risk factors estimated from the agent’s described capabilities.

MAESTRO 7-layer threat model

Per-layer threats for this agent. Layers tagged “not certain from listing” are general, caveated commentary where the public description didn’t pin that layer.

L1 · Foundation Models✓ mapped

Uses FLUX.2 Klein 4B with 1-bit/Ternary quantization. Primary threats include adversarial prompt injection to bypass safety filters, model evasion, and the generation of misaligned, biased, or harmful visual outputs.

L2 · Data Operations⚠ not certain from listing

Not certain from the listing — details regarding the training dataset, fine-tuning data, or quantization calibration datasets are not provided, leaving potential risks around data lineage, copyright infringement, and training data poisoning unaddressed.

L3 · Agent Frameworks⚠ not certain from listing

Not certain from the listing — there is no indication of an agentic orchestration framework, planning capabilities, or tool-use integration, suggesting the system operates as a direct, single-step inference pipeline.

L4 · Deployment & Infrastructure✓ mapped

Features WebGPU acceleration for client-side execution and open weights on Hugging Face. Key threats include WebGPU-based browser exploits, local resource exhaustion, and the risk of supply-chain attacks if malicious model weights are distributed.

L5 · Evaluation & Observability⚠ not certain from listing

Not certain from the listing — no built-in evaluation, prompt filtering, or output monitoring mechanisms are mentioned, which could allow the unchecked generation of unsafe or policy-violating imagery.

L6 · Security & Compliance (cross-cutting)⚠ not certain from listing

Not certain from the listing — while licensed under Apache 2.0, there are no details regarding compliance with generative AI safety standards, copyright protection mechanisms, or user data privacy policies.

L7 · Agent Ecosystem⚠ not certain from listing

Not certain from the listing — the agent operates as a standalone horizontal utility with no described multi-agent coordination, marketplace integrations, or ecosystem dependencies.

MAESTRO — the 7-layer agentic threat-modeling framework (Cloud Security Alliance / Ken Huang).