ClawHub
Back to skill

Security audit

Venture Delegation

Security checks for vulnerabilities and agentic risk

Overview

This is an instruction-only planning skill whose delegation, eval-command, and local note-writing behavior is visible and aligned with its stated purpose.

Install this only if you want an agent to turn broad planning requests into delegated work plans. Review generated eval commands before running them, and avoid using it with confidential tasks unless you are comfortable with local plan/log files and possible sub-agent or model-provider routing of task context.

SkillSpector

Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • 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 (3)

Vague Triggers

Medium
Confidence
84% confidence
Finding
The skill description is extremely broad and explicitly triggers on generic phrases like 'break this down' or 'analyze this opportunity,' which can cause the skill to activate in many ordinary conversations. That creates a real risk of unintended delegation behavior, including unnecessary planning, model routing, and downstream file-writing or orchestration in contexts where the user did not clearly request this workflow.

Missing User Warnings

Low
Confidence
80% confidence
Finding
The instruction to write the full plan to `workspace/DELEGATION_PLAN.md` introduces file-writing side effects without any user-facing notice or confirmation step. In a broadly invoked planning skill, this can unexpectedly persist potentially sensitive project details to disk and may surprise users who expected analysis only.

Missing User Warnings

Low
Confidence
88% confidence
Finding
Appending timing and pass/fail data to `.learnings/LEARNINGS.md` creates persistent local logging without notice, which can accumulate sensitive operational metadata over time. Even if intended for harmless self-improvement, hidden retention of execution history can expose project names, workflow details, and behavioral traces beyond the user's expectations.

Static analysis

No suspicious patterns detected.