ClawHub
Back to skill

Security audit

Critical Thinking

Security checks for vulnerabilities and agentic risk

Overview

This skill clearly installs a user-approved decision-quality rule into a workspace instruction file and does not hide broader access or execution behavior.

Install only if you want this workspace's agent behavior to include the displayed decision-quality rule. Review the exact AGENTS.md path and inserted block before approving, and remember it applies only in workspaces that load that file.

A.I.G

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

SkillSpector

Vulnerability Patterns
  • Memory PoisoningPersistent Context Injection, Context Window Stuffing, Memory Manipulation
  • 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)

Persistent Context Injection

Medium
Category
Memory Poisoning
Content
- the exact target path;
   - the exact rule block below;
   - whether the action will insert, replace an older equivalent block, or do nothing.
7. Obtain explicit approval before editing the persistent instruction file.
8. Preserve all unrelated instructions. Insert the block in the most relevant general-rules section; otherwise append it cleanly.
9. Re-read the file and verify that the block appears exactly once.
10. Report the changed path and explain that the behavior applies only when that AGENTS.md is actually loaded. Recommend verification in a fresh session when possible.
Confidence
90% confidence
Finding
This skill is explicitly designed to modify a persistent instruction file (AGENTS.md) so future agent behavior changes beyond the current interaction. Although it requires explicit user approval and includes some safety constraints, it still performs durable context injection, which can be abused to alter agent policy, create hidden persistence, or normalize behavior changes that outlive the initiating request.

Static analysis

No suspicious patterns detected.