Security audit

crucible

Security checks for vulnerabilities and agentic risk

Overview

This skill is a disclosed Chinese-language software delivery workflow that can write project files and run reviews, but I found no hidden persistence, credential handling, exfiltration, or remote-code behavior.

Install this only if you want a substantial Chinese-language, multi-stage development workflow that may modify code and create documentation/tests. For small edits, invoke it deliberately or avoid auto-use; also check the command path after installation because the packaged command points to a hardcoded Windows Administrator location.

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

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
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (4)

Vague Triggers

Medium

Confidence: 91% confidence
Finding: The trigger description contains very broad activation phrases such as 'building something from scratch', 'full process', and 'team-style development', which can match many ordinary user requests. This can cause the skill to activate unexpectedly and steer the agent into a heavyweight orchestration flow the user did not explicitly request, increasing the chance of prompt-scope confusion and unintended behavior.

Natural-Language Policy Violations

Medium

Confidence: 88% confidence
Finding: The skill metadata and instructions are written to operate in Chinese without indicating any user-language negotiation or fallback. This can lead to mismatches between user intent and agent output, causing misunderstanding of requirements, review results, or safety-relevant guidance in multilingual environments.

Vague Triggers

Medium

Confidence: 79% confidence
Finding: The trigger description is overly broad, making the skill likely to activate for many ordinary software requests. Because this skill can orchestrate multi-stage agent execution, broad activation increases the chance of unintended delegation, context overreach, and unnecessary exposure of user prompts or repository context to multiple sub-agents, which expands the attack surface for prompt injection and unsafe automation.

Natural-Language Policy Violations

Medium

Confidence: 95% confidence
Finding: The file is written entirely in Chinese and gives process instructions without offering any language choice or documenting that Chinese is a required locale. In a multi-user or mixed-language environment, this can cause users or downstream agents to misunderstand requirements, reviews, or constraints, which may indirectly weaken security and delivery quality.

Static analysis

No suspicious patterns detected.