proxymock

Security checks across malware telemetry and agentic risk

Overview

This skill is a coherent helper for proxymock, but it can capture, store, replay, and optionally upload real API or database traffic.

Install this only if you intend to use proxymock and already trust the proxymock binary. Use local or staging traffic when possible, avoid recording production secrets or regulated data, review and redact RRPair files before sharing or uploading, unset proxy environment variables after use, and confirm replay targets before running load or cloud workflows.

SkillSpector

By NVIDIA

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access

Findings (6)

Lp3

Medium

Category: MCP Least Privilege
Confidence: 92% confidence
Finding: The skill clearly enables network-impacting operations such as traffic capture, proxying, replay against target URLs, cloud push/pull, and MCP server startup, yet only declares a binary requirement and no corresponding permissions or safety constraints. This can cause the agent to invoke a capability that intercepts or transmits sensitive traffic without explicit user awareness or policy gating.

Vague Triggers

Medium

Confidence: 82% confidence
Finding: The activation description is very broad, covering generic testing, CI, traffic replay, mocking, and any task involving proxymock or Speedscale. Such broad routing can cause the skill to be selected in situations where the user did not intend network interception, replay, or cloud-related actions, increasing the chance of accidental sensitive-data handling or disruptive test execution.

Missing User Warnings

High

Confidence: 96% confidence
Finding: The skill describes recording live API/database traffic and pushing snapshots to cloud services, but does not warn that captured traffic may include credentials, tokens, cookies, PII, database contents, or internal endpoints. In this context, the omission is especially dangerous because the tool's primary purpose is to intercept and store real traffic, making inadvertent collection and exfiltration of sensitive data highly plausible.

Missing User Warnings

Medium

Confidence: 94% confidence
Finding: The proxy environment instructions tell users to export http_proxy, https_proxy, grpc_proxy, and all_proxy without clearly warning that this reroutes application traffic through a local interception proxy. That omission can lead users to unintentionally capture or alter unrelated service traffic, including authentication flows and database connections, especially if the environment variables persist in a shell or CI environment.

Missing User Warnings

Medium

Confidence: 90% confidence
Finding: The CLI reference explicitly documents recording traffic and redirecting app/proxy logs to disk, but provides no warning that captured requests, responses, headers, credentials, tokens, cookies, or database payloads may contain sensitive data. In a traffic-capture tool, omission of privacy and secret-handling guidance materially increases the risk of inadvertent collection and persistence of regulated or confidential data.

Missing User Warnings

Medium

Confidence: 86% confidence
Finding: The Replay section documents writing replay results and observed traffic to disk, again without warning that the stored artifacts can include sensitive request/response content or identifiers from prior captures. Because this skill is specifically for recording, mocking, and replaying real traffic, users are more likely to handle authentic production-like data, which makes silent persistence more dangerous.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal