Claude Code Delegate

• Outcome: Review
• Latest audit: May 29, 2026 at 9:22 PM UTC
• Version:
  0.1.1

Overview

This skill openly delegates coding to Claude Code, but its default setup can give a background coding agent broad filesystem power without mandatory guardrails.

Install only if you intentionally want an OpenClaw agent to hand coding tasks to a local Claude Code CLI. Before use, set up an enforceable write-guard or path allowlist, run it only inside an isolated project directory, avoid repositories containing secrets, and treat session resume as project-specific rather than reusable across unrelated work.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access
• Supply Chain: Unpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (1)

Vague Triggers

Medium

• Confidence: 94% confidence
• Finding: The skill auto-triggers on extremely broad programming-related requests and then delegates them to a local CLI configured with `--permission-mode bypassPermissions`, which can perform filesystem modifications without interactive approval. This creates a real risk of unintended execution from ordinary user prompts, especially because the skill is user-invocable and designed to run shell commands against a local project directory.

VirusTotal

64/64 vendors flagged this skill as clean.