Back to skill

Security audit

GitLab Agent Self Improvement

Security checks for vulnerabilities and agentic risk

Overview

ClawScan rates this skill suspicious because a broadly described self-improvement workflow is reported to make GitLab project changes and may be triggered without clear project-specific intent or approval.

Consider installing only if you specifically want a GitLab project-maintenance workflow and understand its potential to publish findings or close merge requests. The scanners report complete inspection of the single Markdown component, but the suspicious verdict reflects meaningful concerns about broad activation and remote changes; it does not establish malicious intent.

Vulnerability Patterns
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
Vulnerability Patterns
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • 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
Findings (3)

The activation description is overly broad and does not distinguish general self-improvement requests from GitLab project work.

Medium
Category
Trigger Abuse
Confidence
96% confidence
Finding

The generic personal-growth description can cause this skill to be selected for requests unrelated to GitLab maintenance. This becomes security-relevant because the selected workflow instructs authenticated remote writes rather than merely providing self-improvement advice.

Content

No source excerpt is available for this finding.

A skill described as personal growth directs changes to a remote GitLab project's merge requests.

Medium
Category
Excessive Agency
Confidence
96% confidence
Finding

A skill advertised as personal self-improvement directs the agent to publish findings and modify merge-request state in a fixed remote project. If followed with available GitLab permissions, these instructions can turn a general reflection request into unauthorized external writes; the explicit project reference clarifies the destination but does not establish user authorization.

Content

No source excerpt is available for this finding.

The workflow sends findings to GitLab and closes existing merge requests without warning about external disclosure or workflow changes.

Medium
Category
Output Handling
Confidence
95% confidence
Finding

The instruction to write findings into a remote merge request lacks a check for sensitive information or approval to disclose it. The age-based closure rule also permits changes to existing review workflows without confirming that closure is appropriate; limiting it to self-created merge requests reduces scope but does not establish authorization.

Content

No source excerpt is available for this finding.