Security audit

Kimi Use

Security checks for vulnerabilities and agentic risk

Overview

This skill is a straightforward Kimi API helper, but users should know that prompts and selected local image files are sent to the configured Kimi endpoint.

Install only if you are comfortable sending your prompts, translation text, and any chosen local image files to Kimi or the KIMI_API_HOST you configure. Do not pass secrets, private documents, or regulated data as prompts or image paths, and use only a trusted API host.

SkillSpector

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
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 (4)

Missing User Warnings

Medium

Confidence: 92% confidence
Finding: The documentation shows that user prompts and images, including local files encoded as base64, are sent to Kimi's remote API, but it does not explicitly warn users about external data transfer, retention, or privacy implications. This can lead users to unknowingly transmit sensitive text, images, or metadata to a third-party service.

Missing User Warnings

Medium

Confidence: 89% confidence
Finding: The image command reads an arbitrary local file path and transmits its contents to a remote API as base64 without any explicit warning, confirmation, or strong validation that the file is actually an image. In a CLI skill context, this can lead to accidental exfiltration of sensitive local files if a user supplies the wrong path or if non-image files are mislabeled and sent off-host.

External Transmission

Medium

Category: Data Exfiltration
Content: ```bash export KIMI_API_KEY="sk-kimi-xxxx" # 必填（Kimi For Coding 格式） export KIMI_API_HOST="https://api.kimi.com/coding" # Kimi For Coding 端点 export KIMI_MODEL="kimi-for-coding" # 可选，默认 kimi-for-coding export KIMI_VISION_MODEL="kimi-vl-flash" # 可选，默认 kimi-vl-flash ```
Confidence: 89% confidence
Finding: The environment configuration explicitly points the skill at an external API endpoint, meaning prompts and possibly attached content will leave the local environment. In context, this is expected functionality for an AI-backed skill, but it is still security-relevant because users may disclose sensitive data without realizing it is sent off-host.

External Transmission

Medium

Category: Data Exfiltration
Content: ## API 详情 - **API 地址**: `https://api.kimi.com/coding/v1` - **模型**: kimi-for-coding, kimi-vl-flash (视觉) - **视觉**: 支持本地 base64 图片和 URL - **兼容**: OpenAI SDK 风格
Confidence: 90% confidence
Finding: The API details section reiterates that the skill uses Kimi's remote API and supports sending local images or URLs for vision tasks. This increases the chance of inadvertent external transmission of sensitive content if users assume processing is local.

Static analysis

Detected: suspicious.env_credential_access

Environment variable access combined with network send.

Critical

Code: suspicious.env_credential_access
Location: scripts/index.js:19