ClawHub
Back to skill

Security audit

gitlab-work-stats

Security checks for vulnerabilities and agentic risk

Overview

This skill is a disclosed, read-only GitLab work-statistics reporter, but it should only be used with an authorized, least-privilege GitLab account because it reads sensitive activity data.

Install only if you are authorized to inspect the target GitLab activity. Use a dedicated least-privilege SSH/database account, keep server-config.json private, prefer API or parameterized database access where possible, and avoid enabling the optional repository git-log step unless repository metadata access is explicitly permitted.

SkillSpector

Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (1)

Context-Inappropriate Capability

Medium
Confidence
89% confidence
Finding
The documented workflow expands from SQL-based read-only reporting into direct shell execution of Git against on-disk repositories over SSH. Even though it is framed as read-only and warns against sudo, it still instructs an agent to access repository filesystem paths and interpolate user-controlled values such as username and dates into a shell command, broadening the capability surface beyond the skill’s stated purpose and increasing the risk of unauthorized repository metadata access or command injection in downstream implementations.

Static analysis

No suspicious patterns detected.