Back to skill

Security audit

3D模型审计技能

Security checks for vulnerabilities and agentic risk

Overview

The skill appears intended for 3D model auditing, but it overstates its audit coverage and includes fragile parsers that could mislead or disrupt users handling untrusted model files.

Install only if you treat this as an advisory, partial 3D-model checklist rather than a complete legal, content-safety, or printer-readiness gate. Run it on untrusted files with resource limits, manually review copyright/sensitive-content conclusions, and do not run publish.sh unless you intend to publish the skill to ClawHub.

Vulnerability Patterns
  • Insecure Skill Coding PracticesFinds exploitable flaws such as hardcoded secrets or command injection
  • 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 (5)

T09 · Insecure Skill Coding Practices

Warning
Location
scripts/validate_3d_model.py:89
Finding

Partial Binary STL Parsing Allows Dimension Validation Bypass

Content
View full analysis
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Warning
Location
scripts/validate_metadata.py:68
Finding

Unbounded 3MF Decompression and XML Parsing Can Exhaust Resources

Content
View full analysis
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Warning
Location
scripts/validate_3d_model.py:69
Finding

Unbounded ASCII STL and OBJ Parsing Can Cause Denial of Service

Content
View full analysis
= 4: x = float(parts[1]) y = float(parts[2]) z = float(parts[3]) vertices.append((x, y, z)) elif line.startswith('f '): # 面 face_count += 1 ``` ### Technical Analysis The ASCII STL fallback reads the entire file into memory. It then uses `re.findall`, which creates a list containing every matched coordinate tuple, and subsequently creates another list containing converted floating-point tuples. The OBJ parser reads incrementally but accumulates every vertex in memory without a file-size or vertex-count limit. Bounding-box calculation later creates three additional lists for all X, Y, and Z values. Because model files are untrusted audit inputs, an attacker can intentionally provide a very large ASCII STL or OBJ file to trigger excessive memory allocation and CPU work. ### Attack Path 1. An attacker creates an oversiz ...[truncated 893 chars]
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Warning
Location
scripts/validate_metadata.py:70
Finding

Ambiguous 3MF Model-Part Selection Allows Metadata Spoofing

Content
View full analysis
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Note
Location
scripts/validate_3d_model.py:146
Finding

OBJ Material References Can Escape the Model Directory and Probe File Existence

Content
View full analysis
Remediation
View remediation
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Findings (18)

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The skill is presented as a general 3D compliance auditor, but the finding indicates it mainly performs 3MF metadata-required-field checks while lacking the claimed format validation, infringement analysis, sensitive-content review, and print-size assessment. In a compliance workflow, such overclaiming materially increases risk because incomplete checks may be mistaken for full audit coverage.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

The skill is presented as a general 3D compliance auditor, but the finding indicates it mainly performs 3MF metadata-required-field checks while lacking the claimed format validation, infringement analysis, sensitive-content review, and print-size assessment. In a compliance workflow, such overclaiming materially increases risk because incomplete checks may be mistaken for full audit coverage.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

The skill is presented as a general 3D compliance auditor, but the finding indicates it mainly performs 3MF metadata-required-field checks while lacking the claimed format validation, infringement analysis, sensitive-content review, and print-size assessment. In a compliance workflow, such overclaiming materially increases risk because incomplete checks may be mistaken for full audit coverage.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
Not specified by scanner
Confidence
98% confidence
Finding

该技能清单声明用途是审计STL、OBJ、3MF等3D模型的完整性、版权、敏感内容和打印尺寸,但此文件实际执行的是将当前目录作为Skill发布到clawhub的操作。发布/分发技能并不是3D模型审计功能的明显实现细节,而是与声明用途不同的独立行为。

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
Not specified by scanner
Confidence
99% confidence
Finding

The skill description promises copyright/IP detection, political-sensitive content detection, and inappropriate-content review, but the code only performs basic file-format parsing and bounding-box extraction. This is dangerous because operators may trust the skill to enforce content policy when it has no such detection logic, enabling prohibited models to be falsely treated as screened and approved.

Content

No source excerpt is available for this finding.

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
73% confidence
Finding

The skill appears to require file-reading capability to inspect model files and metadata, but it does not explicitly declare any tool scope or permission boundaries. Missing scope declarations can cause the runtime to grant broader-than-necessary access or make review of the skill's actual capabilities harder, violating least-privilege expectations.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

An activation scope phrased as applicable to 'any 3D model audit scenario' is overly broad and can cause the skill to trigger in contexts where its checks are incomplete or inappropriate. Overbroad routing increases the chance of misuse and overreliance, especially given the other capability mismatches already identified.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The skill name and description are presented entirely in Chinese and frame the skill as a Chinese-language capability, with no indication that users may choose another language or locale. Under the stated policy, forcing a specific language without user opt-in is a natural-language policy violation.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The display name, description, and changelog are written entirely in Chinese, which effectively forces a specific language for users reading the skill metadata. The file does not indicate that the skill is region-specific or provide any user opt-in or alternative locale, so this is a natural-language policy concern under the language/locale rule.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

This code performs a subprocess-style publication action via clawhub publish, which can have external side effects, but the script provides no confirmation prompt or explicit warning about publishing before execution. The generic echo "正在发布..." announces progress, but it does not disclose the implications or give the user a chance to abort.

Content

No source excerpt is available for this finding.

Ae4

Medium
Category
analysis-evasion
Confidence
80% confidence
Finding

Suspicious Unicode normalization or mixed-script content

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The entire skill reference is written only in Chinese and does not indicate that language selection is optional or that the file is intentionally limited to a Chinese-language or China-specific deployment. Under the policy, forcing a specific language or locale without opt-in is a natural-language policy violation.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

This code file contains natural-language documentation and runtime output entirely in Chinese, including the module docstring and later user-facing messages. The policy only allows a fixed language when the skill offers a language choice, opt-in, or a clearly justified locale constraint, which is not present here.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
98% confidence
Finding

The skill metadata claims support for 3MF auditing, but the implementation explicitly rejects 3MF files as unsupported. In a compliance or moderation workflow, this creates a silent coverage gap where users may rely on checks that never occur, allowing disallowed or unsafe 3MF content to pass unreviewed or be skipped entirely.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The manifest describes a compliance skill for STL/OBJ/3MF files that should verify file integrity/format validity, detect copyright infringement, identify political or inappropriate content, and check model size against printer bounds. This file's documented and implemented behavior is limited to checking whether Title, Description, and License metadata exist, with no code for content/IP analysis, geometry inspection, or printer-volume checks.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
95% confidence
Finding

The script's natural-language comments and console output are written in Chinese, which imposes a specific language experience without any opt-in or indication that the skill is intentionally limited to Chinese-speaking users. Under the language/locale policy, locale-specific behavior should either offer a choice or clearly document and justify the constraint.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Low
Category
Not specified by scanner
Confidence
80% confidence
Finding

The file documentation says it validates required fields in 3MF files, while the main validation flow also accepts .stl and .obj extensions. For STL and OBJ, the implementation immediately reports that embedded metadata is unsupported rather than actually validating equivalent metadata sources, creating a mismatch between the apparent multi-format validator behavior and the documented 3MF-only purpose.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
92% confidence
Finding

This code file contains its primary docstring and user-facing status/error messaging in Chinese, and there is no indication that language selection is optional or region-specific. The policy for natural-language content requires avoiding a forced language or locale unless the constraint is documented and justified.

Content

No source excerpt is available for this finding.