Back to skill

Security audit

Git Tool

Security checks for vulnerabilities and agentic risk

Overview

This is a straightforward Git helper skill whose state-changing behavior fits its stated repository-management purpose, though users should treat its Git commands as real repository mutations.

Install only if you want an agent to run real Git operations in your repositories. Review status and diffs before commit, stash, pull, or push, and be aware that commit stages files by default when no file list is supplied.

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
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Behavioral ASTexec() Call, eval() Call, Dynamic Import
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Findings (6)

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
94% confidence
Finding

The skill advertises shell-backed functionality but does not declare any explicit tool scope such as allowed-tools or permissions. This increases the chance an agent can invoke shell operations more broadly than intended, which is risky for a repository-management skill that can modify local state and potentially interact with configured Git remotes.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The activation text is broad enough to trigger this skill for general repository-management or automation tasks, which may cause an agent to select it in situations involving sensitive repositories or destructive Git actions. Overbroad routing expands the attack surface because the skill includes state-changing commands like init, branch, stash, and commit.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The documentation presents mutating Git commands without any warnings, confirmations, or discussion of side effects such as altering repository state, creating commits, or stashing local work. In an agent context, omission of these guardrails makes unintended changes more likely, especially when users ask for broad automation help.

Content

No source excerpt is available for this finding.

subprocess module call

Medium
Category
Dangerous Code Execution
Confidence
70% confidence
Finding

subprocess module calls execute external commands. Without careful input validation, this enables command injection.

Content

Scanner excerpt · scripts/git.py (reported line 16)May include surrounding context.

python
"""Run git command."""
    cmd = ['git'] + args
    try:
        result = subprocess.run(cmd, capture_output=capture, text=True)
        if result.returncode == 0:
            if capture:
                print(result.stdout)

Missing User Warnings

Low
Category
Not specified by scanner
Confidence
84% confidence
Finding

The code performs a network operation via git push, which can transmit local repository contents to a remote server. Although the function name suggests its purpose, there is no confirmation prompt or explicit user disclosure in code about the outbound data transfer.

Content

No source excerpt is available for this finding.

Missing User Warnings

Low
Category
Not specified by scanner
Confidence
76% confidence
Finding

git pull contacts a remote and updates local repository state, which can affect the user's working tree. The script does not provide any explicit warning, confirmation, or detailed help text describing this behavior.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.