ClawHub
Back to skill

Security audit

Jira PAT 管理基础版

Security checks for vulnerabilities and agentic risk

Overview

The skill is a Jira PAT helper, but it can make live Jira changes while its trigger, data-flow disclosure, and write-safety guidance are too broad or unclear.

Review before installing. Use a least-privilege Jira PAT, point it only at the intended Jira instance, and treat status changes, comments, field updates, and issue creation as live production actions. Avoid using this skill as a general reporting or analytics helper unless the task is explicitly Jira-related.

SkillSpector

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

Intent-Code Divergence

High
Confidence
94% confidence
Finding
The skill claims all data stays local and is not uploaded to the cloud, yet it explicitly instructs the agent to send Jira data over the network to an external/self-hosted Jira API. This can mislead users about where sensitive issue contents, metadata, and credentials are transmitted, causing unsafe use in regulated or confidential environments.

Vague Triggers

High
Confidence
91% confidence
Finding
The trigger condition is overly broad and mismatched to the skill's Jira-specific scope, inviting activation for generic data analysis, reporting, or visualization tasks unrelated to Jira. In an agent environment with exec/network access and write capability declared in metadata, this increases the chance of the skill being selected inappropriately and making live changes or external requests in the wrong context.

Missing User Warnings

Medium
Confidence
93% confidence
Finding
The skill advertises mutating capabilities such as status changes, comments, field updates, and issue creation without a prominent warning that these actions affect live Jira data. Users may assume the skill is informational, causing accidental production changes, workflow disruption, or unintended disclosure through comments and field edits.

Static analysis

No suspicious patterns detected.