Security audit

Claw Go

Security checks for vulnerabilities and agentic risk

Overview

This text-only game skill is mostly coherent, but it asks to reuse broad long-term memory and infer user profile details without clear consent or scoping.

Review this skill before installing if you do not want a game to use general long-term memory or infer personal preferences from past chats. It appears text-only and does not run code or publish externally, but users should prefer a version that limits personalization to explicit Claw Go state and gives clear opt-in, correction, and deletion controls for saved preferences.

A.I.G

Vulnerability Patterns

Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
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)

T05 · Unauthorized Access and Privilege Escalation

Warning

Location: references/game-design.md:45
Finding: Overbroad Long-Term Memory Access and User Profiling
Content: View full analysis
Remediation: View remediation

SkillSpector

Vulnerability Patterns

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

Findings (7)

Natural-Language Policy Violations

Medium

Category: Not specified by scanner
Confidence: 81% confidence
Finding: The manifest uses natural-language keys such as "得意虾" and "震惊虾" as fixed expression labels, which imposes a specific language in the skill data without any visible opt-in or documented locale limitation. Under the language/locale policy, forcing one language without user choice or justification can be a policy concern.
Content: No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium

Category: Not specified by scanner
Confidence: 88% confidence
Finding: The premium upsell and fallback templates are written only in Chinese, while the file otherwise indicates language should depend on user_language. This can lead to Chinese-language output being used by default for users who have not explicitly opted into that locale.
Content: No source excerpt is available for this finding.

Missing User Warnings

Medium

Category: Not specified by scanner
Confidence: 92% confidence
Finding: The contract defines optional memory tag syncing tied to a user_id and owner interactions, but provides no privacy notice, consent model, retention guidance, or limits on what tags may be stored. In a consumer game skill, this can lead to silent collection of preference/profile data and misuse or over-collection of behavioral information.
Content: No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium

Category: Not specified by scanner
Confidence: 88% confidence
Finding: The examples for image generation and TTS use Chinese strings for user-facing content, but the document does not say this is locale-specific or optional. This can indicate a language policy issue because the skill appears to force a specific language without documenting user choice or a justified regional constraint.
Content: No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium

Category: Not specified by scanner
Confidence: 82% confidence
Finding: Line L130 specifies an English mapping only when user_language=en, while the surrounding design centers Chinese labels and examples as the default. This creates a language policy concern because it constrains output language based on inferred state rather than explicit user choice, and does not offer other locale options or opt-in behavior.
Content: No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium

Category: Not specified by scanner
Confidence: 90% confidence
Finding: Inferring and persisting user_language from recent messages means the system derives and stores user attributes without explicit consent or confirmation. While low severity in this game context, it can still create privacy issues, incorrect profiling, and unwanted retention of inferred personal preferences, especially for multilingual users or shared accounts.
Content: No source excerpt is available for this finding.

Natural-Language Policy Violations

Low

Category: Not specified by scanner
Confidence: 93% confidence
Finding: The instruction says to 'match the user's language when clear, otherwise default to Chinese.' This creates a language-policy issue because it forces a specific language in ambiguous cases rather than asking the user to choose or offering options.
Content: No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.