Security audit

Code Dev Toolkit

Security checks for vulnerabilities and agentic risk

Overview

This code workflow skill is mostly coherent, but it asks for command execution and describes CI/CD release and webhook-token use without enough scoping or confirmation safeguards.

Review this before installing in sensitive repositories. Use it only with explicit project, branch, and environment targets; require confirmation before any delivery or release step; keep auto-delivery disabled; and use least-privilege CI tokens and approved webhook endpoints.

SkillSpector

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
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access

Findings (4)

Vague Triggers

Medium

Confidence: 90% confidence
Finding: The skill uses broad natural-language trigger phrases like '请编排以下开发任务' that can match ordinary user requests without clear activation boundaries. In an agent with exec capability, ambiguous invocation can cause unintended orchestration, code review, or shell-backed actions to run on the wrong project or context.

Vague Triggers

Medium

Confidence: 91% confidence
Finding: The CI/CD examples describe high-level execution requests such as running a full delivery pipeline, automated review, and delivery, but do not constrain repository, branch, environment, or approval conditions. In a skill exposing exec and potential network use, this raises the risk of unintended pipeline execution, review actions, or release-related side effects from loosely phrased prompts.

Missing User Warnings

Medium

Confidence: 88% confidence
Finding: The skill advertises exec capability and later provides setup steps that create directories and write configuration files, but it does not clearly warn users that invoking the skill may modify the local filesystem or run shell commands. This can lead to unintended system changes, especially when the skill is used in automated agent contexts where users may expect analysis rather than mutation.

Missing User Warnings

Medium

Confidence: 86% confidence
Finding: The skill includes network-capable CI/CD integration and webhook credential configuration but does not clearly warn that data, metadata, or code review artifacts may be transmitted to external services. This creates a privacy and secret-handling risk, particularly in enterprise codebases where repository details, audit logs, or tokens may be sensitive.

Static analysis

No suspicious patterns detected.