ClawHub
Back to skill

Security audit

需求调度官

Security checks for vulnerabilities and agentic risk

Overview

This skill is a disclosed Chinese requirements-management workflow that can launch implementation agents, run tests, and merge code only as part of explicit dispatch and acceptance commands.

Install only in repositories where you are comfortable letting the skill create planning files, launch implementation subagents, run the project test suite, and merge worktree branches after review. Review the dispatch and accept steps carefully, especially in untrusted repositories whose tests or build scripts may execute arbitrary code.

SkillSpector

Vulnerability Patterns
  • 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
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Findings (1)

Missing User Warnings

Medium
Confidence
96% confidence
Finding
The skill’s dispatch and accept flows explicitly spawn background subagents that develop code, run tests, and merge worktree branches, but the skill description and entrypoint guidance do not clearly warn users that invoking these commands will execute repository code and tests. That omission can lead to uninformed consent and accidental execution of untrusted project code or agent-driven changes, which is especially risky in repositories containing unsafe test hooks, build scripts, or developer tooling.

Static analysis

No suspicious patterns detected.