ClawHub
Back to skill

Security audit

AI Code Review

Security checks for vulnerabilities and agentic risk

Overview

The skill mostly describes a disclosed AI code-review workflow, but it also includes under-scoped service auto-restart and skill-publishing capabilities that exceed ordinary review behavior.

Review carefully before installing. The LLM, Whisper, and Discord features mean diffs, audio, and notifications may leave your environment. The larger concern is that the skill also documents restarting host/container services and publishing skills; use it only where those actions are explicitly intended, scoped, and manually controlled.

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
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (5)

Context-Inappropriate Capability

High
Category
Not specified by scanner
Confidence
97% confidence
Finding

The health-check script's monitoring, auto-restart, and alerting behavior is materially outside the stated code-review purpose and implies operational control over host or container services. Scope expansion like this is dangerous because users may install a review skill expecting analysis-only behavior, while it also gains the ability to inspect system state and restart services, increasing the risk of abuse or unintended disruption.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

This markdown file states that the skill performs LLM analysis, voice-note transcription, and Discord webhook notifications, all of which may send code or audio-derived content to third-party services. The description does not include any user-facing warning about privacy, data sharing, or the fact that content may leave the local system.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The skill description states that diffs are sent to GPT-4o and recordings are transcribed via Whisper, but it does not provide an explicit privacy/data-handling warning to users. This is dangerous because code diffs and voice notes may contain sensitive source code, secrets, internal discussions, or regulated data that are transmitted to third-party services without clear informed consent.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The documented publish_skill() capability is outside the core code-review function and introduces a software supply-chain publishing action from within the same skill. Even if not inherently malicious, bundling publication into a review skill increases blast radius: a user invoking code review may unknowingly grant a path to distribute artifacts externally or automate release actions beyond the expected scope.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
50% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · SKILL.md (reported line 64)May include surrounding context.

md
|---|---|---|---|
| `OPENAI_API_KEY` | Yes | — | OpenAI API key for GPT-4o and Whisper |
| `DISCORD_WEBHOOK_URL` | No | — | Discord webhook URL for notifications |
| `CLAWHUB_API_URL` | No | `https://api.clawhub.com/v1` | ClawHub API base URL |

## Health Check

Static analysis

No suspicious patterns detected.