Back to skill

Security audit

Skill Update Litiao

Security checks for vulnerabilities and agentic risk

Overview

This is a disclosed skill-updater with preview, confirmation, backup, and rollback guidance, though users should be careful about the unpinned npx ClawHub commands.

Install only if you want an agent to help manage skill updates. Before following its commands, prefer a pinned or trusted preinstalled ClawHub CLI, review the preview diff, confirm backups were created, and approve updates or migrations only when you understand the changed files and impact.

Vulnerability Patterns
  • Insecure DependenciesIntroduces malicious components through unsafe dependency sources
  • 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
Findings (1)

T08 · Insecure Dependencies

Warning
Location
SKILL.md:35
Finding
Unpinned ClawHub CLI Execution Through npx## Vulnerability Details **File Location**: `SKILL.md:35-36` and `preview.md:7` **Vulnerability Type**: Unpinned third-party package execution **Risk Level**: Medium ### Vulnerable Code `SKILL.md:35-36`: ```bash npx clawhub outdated # List skills with updates npx clawhub info <slug> # Show available versions ``` `preview.md:7`: ```bash npx clawhub fetch <slug>@<version> --dir /tmp/preview-<slug> ``` ### Technical Analysis The documented commands execute `clawhub` through `npx` without pinning the CLI package to a specific, reviewed version. If the package is not already available locally, `npx` may retrieve it from the configured package registry and execute it immediately. Although the fetch command supplies a version for the skill being downloaded, that version does not constrain the version or integrity of the `clawhub` executable itself. The effective CLI implementation can therefore change after this skill has been reviewed. This creates a supply-chain trust boundary in which compromise of the registry account, package distribution infrastructure, package name resolution, or a newly published malicious CLI version could result in arbitrary code execution. ### Attack Path 1. An attacker compromises the package publisher, registry distribution path, or another component controlling the package resolved as `clawhub`. 2. The attacker publishes or serves a malicious version containing install-time or runtime code. 3. A user or Agent follows the skill instructions and runs one of the unpinned `npx clawhub` commands. 4. `npx` resolves and downloads the attacker-controlled package when no trusted local version is available. 5. Package lifecycle logic or the invoked CLI executes with the permissions of the current user or Agent process. 6. The malicious package can access resources available to that process before optionally displaying expected CLI output to conceal t ...[truncated 627 chars]
Remediation
## Remediation Suggestions 1. Pin the ClawHub CLI to an exact, audited version rather than relying on the registry's current version: ```bash npx --yes clawhub@<verified-version> outdated npx --yes clawhub@<verified-version> info <slug> npx --yes clawhub@<verified-version> fetch <slug>@<version> --dir /tmp/preview-<slug> ``` 2. Prefer a trusted, preinstalled CLI managed through a lockfile containing package integrity hashes. 3. Configure an approved package registry and verify package provenance, signatures, and integrity before installation or execution. 4. Where supported, install the reviewed package in advance and execute it in offline mode so `npx` cannot silently retrieve a newer implementation. 5. Review dependency trees and package lifecycle scripts before approving a new CLI version. 6. Run update tooling with least privilege and restrict its filesystem, credential, and network access to the minimum required. 7. Update both `SKILL.md` and `preview.md` so every example follows the same pinned and verified execution procedure.
Vulnerability Patterns
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Rogue AgentSelf-Modification, Session Persistence
  • 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
Findings (5)

Self-Modification

High
Category
Rogue Agent
Content
---
name: skill-update-litiao
description: "Safely update skills with preview, migration support, and user validation. Never lose data or break workflows."
---

## Safe Skill Updates
Confidence
85% confidence
Finding
This skill is explicitly designed to update skills, which is a form of self-modification of agent capabilities and instructions. Even though the document includes safety steps like preview, backup, and confirmation, self-update functionality remains high risk because it can alter future behavior, introduce malicious or incompatible instructions, and expand the impact of any compromised source or mistaken approval.

Rp1

Medium
Category
MCP Rug Pull
Confidence
91% confidence
Finding
The skill instructs use of `npx clawhub` without pinning a specific package version, which can cause execution of whatever version is currently resolved from the registry. In a self-update workflow, that increases supply-chain risk because behavior may change unexpectedly or a compromised upstream release could be fetched at execution time.

Rp1

Medium
Category
MCP Rug Pull
Confidence
91% confidence
Finding
The `npx clawhub info <slug>` command also relies on an unpinned package resolution path, meaning the update-checking logic may run different code over time without user awareness. Because this skill is about modifying installed skills, using mutable tooling adds avoidable supply-chain and integrity risk.

Vague Triggers

Medium
Confidence
91% confidence
Finding
The instruction to check for updates whenever the user 'mentions a skill' is overly broad and may trigger update-related behavior during ordinary discussion rather than a clear maintenance request. In a skill that can lead to self-modification, unsolicited update checks can expand the action surface and nudge users toward risky operations they did not intend to start.

Rp1

Medium
Category
MCP Rug Pull
Confidence
95% confidence
Finding
The command uses `npx clawhub` without pinning a specific package version, so execution may resolve to whatever version is currently published or otherwise available at runtime. In a skill-update workflow, this increases supply-chain risk because a compromised or unexpected package version could be fetched and executed during preview, affecting the local environment or producing misleading update data.

Static analysis

No suspicious patterns detected.