ClawHub
Back to skill

Security audit

Weekly Review

Security checks for vulnerabilities and agentic risk

Overview

The skill is a coherent weekly AI-usage review and report renderer; its sensitive review inputs are disclosed, but users should keep data collection scoped.

Install only if you want an agent to help summarize AI usage and conversations into a weekly report. Provide or approve the review data explicitly, keep any disk checks limited to relevant project folders, and do not use the legacy SQLite analyzer unless you intentionally point it at a database.

SkillSpector

Vulnerability Patterns
  • 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
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (3)

Lp3

Medium
Category
MCP Least Privilege
Confidence
85% confidence
Finding
The skill declares executable capabilities and data access patterns (Python CLI use, reading input files, writing reports/charts, and likely environment access through the runtime) without an explicit permissions model. This can cause the host agent to invoke file and execution behaviors that users did not clearly authorize, increasing the chance of unintended data exposure or modification.

Vague Triggers

Medium
Confidence
84% confidence
Finding
The skill advertises activation for broad tasks like prompt improvement, weekly summaries, usage viewing, visualization, and conversation cleanup, which overlap with many common assistant requests. Over-broad routing can cause the agent to load this skill in contexts where its file-access, reporting, or cleanup behavior is unnecessary, leading to scope creep and accidental processing of sensitive conversation data.

Vague Triggers

Medium
Confidence
86% confidence
Finding
The read_when triggers are ambiguous and underspecified, containing generic phrases such as prompt improvement, cleanup of open conversations, and automated weekly reports. Such triggers can match normal productivity requests and silently expand the skill's reach into reviewing sessions, touching files, or recommending cleanup actions without sufficient context or consent.

Static analysis

No suspicious patterns detected.