ClawHub
Back to skill

Security audit

Agent Architecture Analysis

Security checks for vulnerabilities and agentic risk

Overview

This skill is a read-only architecture audit guide that asks the agent to inspect a user-provided codebase and produce evidence-based recommendations.

Install this if you want an agent to perform a structured architecture review of an LLM or agentic codebase. Expect it to read source files under the codebase path you provide and run search commands for evidence; review any generated recommendations before applying code changes.

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
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (2)

Unvalidated Output Injection

High
Category
Output Handling
Content
| **Weak** | Tools return arbitrary dicts, no validation layer |

**Anti-patterns:**
- Tool handlers that directly execute LLM output
- `eval()` or `exec()` on LLM-generated code
- No separation between decision (LLM) and execution (code)
- Magic method dispatch based on string matching
Confidence
80% confidence
Finding
Model output is used without validation or sanitization. Unvalidated output injected into downstream contexts (SQL, shell, HTML) enables injection attacks and arbitrary code execution.

Vague Triggers

Medium
Confidence
89% confidence
Finding
This manifest description starts with several specific trigger phrases, but then broadens activation to 'also applies when comparing frameworks or planning agent improvements.' Those conditions are common, high-level activities that could overlap with many general architecture or planning requests, making invocation scope less precise.

Static analysis

No suspicious patterns detected.