For the complete documentation index, see llms.txt. This page is also available as Markdown.

Skill Details

About skill details

You can investigate an individual skill to understand how it is used in your organization, review its security assessment, examine the evidence behind identified findings, and determine whether remediation is required.

Explore

The Explore tab provides an overview of the skill, including its security assessment, description, findings, capabilities, usage statistics, and available metadata.

Description

Describes the skill and its behavior.

Security assessment

For skills classified as Dangerous, the Explore tab includes two additional sections:

  • Why This Verdict: Explains the malicious behavior identified by OX and provides supporting evidence.

  • Recommendation: Provides recommended actions to address the identified security risks.

For skills classified as Borderline or Safe, these sections are not displayed. You can still review the skill description, findings, capabilities, and available metadata to understand its behavior and security assessment.

Usage statistics

The Explore tab displays the following information:

Field
Description

Users

Number of users in your organization who have the skill on their devices.

Findings

Number of security findings identified for the skill.

The values appear as clickable counters. Select a counter to open the corresponding Users or Findings tab.

Findings

The Findings section displays the security issues identified in the skill as compact chips.

Hover over a finding to display a tooltip containing:

  • An explanation of why the finding was identified.

  • The relevant code snippet and file containing the finding.

  • Vulnerability category tags associated with the finding.

For example, if a skill is identified as sending secrets over the network, the tooltip explains the behavior and references the relevant code.

To review all identified findings, select View All Findings.

Capabilities

Depending on the skill, VibeSec can highlight potentially risky capabilities, such as:

  • Sending data over the network.

  • Fetching remote content.

Capabilities are displayed as chips. Hover over a capability to display an explanation of why VibeSec identified the behavior, when available.

These capabilities do not necessarily indicate that the skill is malicious or vulnerable. The section highlights capabilities that may represent a security risk and does not provide a complete description of everything the skill can do. If VibeSec does not identify potentially risky capabilities, this section is not displayed.

Additional metadata

Additional metadata can include:

  • Declared tools that the skill is permitted to call.

  • Network hosts with which the skill communicates.

Users

The Users tab shows the users in your organization who have the skill on their devices. If the same skill is present on multiple users' devices, each user is listed separately.

The tab includes the following information:

Column
Description

User

Email address of the user associated with the skill.

Projects

Projects associated with the skill for the user, when available.

First Seen

When VibeSec first detected the skill for the user.

Last Seen

When VibeSec most recently detected the skill for the user.

Findings

The Findings tab displays the security issues identified in the skill and provides supporting evidence to help you understand why the skill received its security assessment.

Findings & Recommendations

The Findings & Recommendations section lists the identified security findings.

Column
Description

Name

Name of the security issue or behavior identified in the skill, such as Hidden characters or Injection phrasing hidden in an HTML comment.

Evidence

Provides access to the code or content that triggered the finding.

Location

File and line number where the finding was identified, when available.

Hover over a finding to display additional information, including:

  • An explanation of the detected behavior and why it was identified as a security finding.

  • The source file containing the finding.

  • A code snippet showing the relevant content, including its line number.

Use the explanation, evidence, and location information to identify the content responsible for the finding and investigate the skill.

Last updated