AgentReadyHomeAgent Listing

← Zhipu Web Search MCP

Zhipu Web Search MCP — agentic threat model

7.5AIVSS 7.5 · High

The Zhipu Web Search MCP presents a moderate risk profile, primarily acting as a vector for indirect prompt injection via untrusted web content and financial exposure through API key theft.

OWASP AIVSS score rationale

AIVSS = (CVSS_Base + AARS) × Mitigation_Factor, where AARS = (10 − CVSS_Base) × (Factor_Sum / 10) × ThM
CVSS base 6.5AARS uplift 1.01Factor sum 2.9/10Threat ×1.0Mitigation ×1.0
Autonomy of Action
0.30
Goal-Driven Planning
0.10
Self-Modification
0.00
Dynamic Tool Use
0.20
Persistent Memory
0.00
Contextual Awareness
0.50
Dynamic Identity
0.10
Multi-Agent Interactions
0.60
Non-Determinism
0.70
Opacity & Reflexivity
0.40

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

The underlying foundation model (Zhipu BigModel) is susceptible to indirect prompt injection payloads embedded in retrieved web content, which can reprogram downstream model behavior.

L2 · Data Operations✓ mapped

Data operations rely on real-time web scraping and ranking. Threats include data poisoning via SEO manipulation or malicious web pages designed to feed toxic or misleading grounding data to the agent.

L3 · Agent Frameworks✓ mapped

As an MCP tool, insecure integration into agent frameworks could allow malicious actors to craft queries that exploit the calling framework or trigger unintended tool execution paths.

L4 · Deployment & Infrastructure✓ mapped

The deployment relies on a Zhipu API key. Compromise of this key leads to direct financial theft via credit exhaustion on the user's BigModel account.

L5 · Evaluation & Observability⚠ not certain from listing

Not certain from the listing — there is no mention of built-in guardrails, content filtering, or logging mechanisms to detect and block malicious search payloads before they reach the model.

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

Not certain from the listing — the directory does not specify authorization policies, access controls for the API key, or compliance certifications for data handling.

L7 · Agent Ecosystem✓ mapped

Designed for the MCP ecosystem, this tool introduces cascading risks where a compromised or malicious agent can abuse the search tool to exfiltrate data or inject payloads into other connected agents.

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