Security audit

智能周报基础版

Security checks across malware telemetry and agentic risk

Overview

The skill is a coherent weekly-report helper, but it under-discloses network/API/callback use while claiming free-version data stays local.

Review this skill carefully before installing. It may be useful for drafting weekly reports, but only use it with non-sensitive work content unless you have confirmed network egress, callback behavior, output directories, caching, and API key storage are disabled or scoped the way you expect.

SkillSpector

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands

Findings (8)

Intent-Code Divergence

High

Confidence: 96% confidence
Finding: The documentation states that FREE-version data is stored locally and not uploaded, yet elsewhere the skill permits external API usage and callback URLs. This creates a misleading privacy and trust boundary: users may provide sensitive work logs or report content believing it remains local when it may be transmitted off-host.

Context-Inappropriate Capability

Medium

Confidence: 86% confidence
Finding: The skill requests exec capability even though its stated purpose is weekly-report drafting and todo generation. Unnecessary shell execution expands the attack surface significantly because prompt-controlled inputs could be transformed into filesystem or command execution actions by the hosting agent.

Description-Behavior Mismatch

Medium

Confidence: 90% confidence
Finding: The skill is presented as a lightweight local FREE tool, but its behavior and examples broaden into networking and asynchronous callbacks. This mismatch weakens informed consent and can cause users or agents to expose internal work context to external services unexpectedly.

Vague Triggers

Medium

Confidence: 82% confidence
Finding: The skill advertises activation through broad natural-language instructions, making accidental or overly permissive invocation more likely. In agent environments, generic triggers can cause the skill to engage on unrelated conversations and operate on logs, files, or outputs without sufficiently specific user intent.

Vague Triggers

Medium

Confidence: 82% confidence
Finding: The usage scenario says the skill can be triggered by generic natural-language instructions without strict scoping. That increases the chance of unintended activation and downstream actions such as reading logs or writing report files based on ambiguous user messages.

Vague Triggers

Medium

Confidence: 80% confidence
Finding: Using a common phrase like '请帮我生成本周周报' as the invocation example encourages broad matching in normal conversation. In an autonomous agent context, everyday wording can lead to unintentional skill activation and hidden file or tool operations without a distinct boundary.

Missing User Warnings

Medium

Confidence: 84% confidence
Finding: The skill describes generating reports and todos into local directories but does not clearly warn users that files will be written. Silent or under-documented file creation can overwrite data, leak sensitive work details into insecure locations, or surprise users operating in shared environments.

Missing User Warnings

Medium

Confidence: 91% confidence
Finding: The documentation acknowledges external API/network use but lacks a clear warning that report content, logs, or context may leave the host. For a work-reporting tool, such data often contains internal project details, so undocumented egress creates meaningful confidentiality risk.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.