Back to skill

Security audit

Ai Intelligent Ar Vr Platform

Security checks for vulnerabilities and agentic risk

Overview

This skill needs Review because it tells users to download, install, and run mutable external code that is not included in the reviewed package.

Review the external repository and dependency list before installing. Prefer a pinned commit, exact dependency versions with hashes, and a virtual environment or container with no unnecessary secrets or host access. Treat the author/organization mismatch and Chinese-only documentation as reasons to verify provenance and terms before use.

Vulnerability Patterns
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • 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
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
Findings (1)

T03 · Remote Payload Retrieval and Execution

Error
Location
SKILL.md:43
Finding

Unpinned Remote Payload Retrieval and Unsafe Dependency Installation

Content
View full analysis

Vulnerability Details

File Location: SKILL.md, lines 43-46
Vulnerability Type: Remote payload retrieval and supply-chain exposure
Risk Level: High

bash
git clone https://github.com/openclaw-skills/ai-intelligent-ar-vr-platform
cd ai-intelligent-ar-vr-platform
pip install -r requirements.txt
python app.py

Technical Analysis

The installation instructions clone an external Git repository and execute its application without pinning the repository to an immutable commit, tag digest, or verified release artifact. The effective application payload is not included in the audited skill package and may therefore change after this package has been reviewed.

The instructions also install the remote repository's requirements.txt without any visible version constraints, cryptographic hashes, signature verification, or trusted-package policy. Python package installation can execute package build and installation hooks, so a malicious or compromised dependency may run code before app.py is launched.

The local skill.json identifies the author as yang1002378395-cmyk, while the installation instructions reference the openclaw-skills GitHub organization. The audited files provide no provenance or ownership explanation connecting these identities. This discrepancy does not prove malicious behavior, but it makes independent source verification especially important.

No malicious payload is embedded in the two audited files. The risk arises from retrieving, installing, and executing mutable code that is outside the reviewed artifact.

Attack Path

  1. An attacker compromises the referenced GitHub repository, gains control of its owner account, or causes users to retrieve a subsequently modified repository state.
  2. Alternatively, the attacker compromises, replaces, or introduces a malicious Python dependency referenced by the remote requirements.txt.
  3. A user follows the documented installation procedure and clones the repository with ...[truncated 1127 chars]
Remediation
View remediation

Remediation Suggestions

  1. Include the complete application source and dependency manifests in the skill package so they can be reviewed together.
  2. If remote retrieval is unavoidable, pin the repository to a specific immutable commit SHA rather than cloning the mutable default branch.
  3. Distribute signed release artifacts and verify their signatures or cryptographic checksums before installation or execution.
  4. Pin every Python dependency to an exact reviewed version and use hashes, such as through a lock file or pip install --require-hashes.
  5. Restrict package installation to approved package indexes and validate package names to reduce dependency-confusion and typosquatting risks.
  6. Disable unnecessary dependency build isolation or installation scripts where feasible, and review packages that require native builds or setup hooks.
  7. Document the relationship between the package author and the referenced GitHub organization so users can verify provenance.
  8. Run retrieved software inside an isolated, least-privilege virtual environment or container with no unnecessary secrets, host mounts, or network access.
  9. Perform a separate security review of the pinned remote source, requirements.txt, resolved dependency tree, and app.py before publishing executable installation instructions.
Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • 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 (3)

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The manifest description and all user-facing markdown content are presented in Chinese only. Under the policy, forcing a specific language without user opt-in is a natural-language policy violation unless the locale restriction is explicitly documented and justified, which is not present here.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The manifest description claims only a generic 'AI intelligent' platform, while the stated skill purpose is an AR/VR application development platform. This mismatch can mislead users, reviewers, or automated routing systems about what the skill actually does, weakening trust boundaries and making it easier for a misleading or repurposed skill to evade appropriate scrutiny.

Content

No source excerpt is available for this finding.

Vague Triggers

Low
Category
Not specified by scanner
Confidence
87% confidence
Finding

The description is too vague to define a clear activation or operational scope, which increases ambiguity about the skill's intended behavior. Overly generic metadata can enable accidental over-broad deployment, user misunderstanding, or concealment of functionality that would otherwise receive closer review.

Content

No source excerpt is available for this finding.