T09 · Insecure Skill Coding Practices
- Location
SKILL.md:74- Finding
Execution of an Unverified Script Outside the Skill Package
- Content
View full analysis
- Remediation
View remediation
Security audit
Security checks for vulnerabilities and agentic risk
The skill is mostly coherent for multi-provider image generation, but it tells users to run an unreviewed proxy script outside the published artifact and persist API keys locally without documented protections.
Install only if you are comfortable with a local proxy that sends prompts and image data to the selected AI provider. Before use, verify the actual proxy and injection scripts from the package you install, avoid putting secrets in prompts, use restricted provider keys with billing limits, and check or delete `~/.openclaw-dev/uni-image-config.json` if you do not want keys retained on disk.
SKILL.md:74Execution of an Unverified Script Outside the Skill Package
SKILL.md:49Persistent Plaintext Storage of Image-Provider API Keys
Referenced artifact was not completely inspected
- `SKILL.md` — This skill manifest
The skill is designed to proxy prompts and image-generation requests to multiple third-party AI providers, but the documentation does not prominently warn users that their prompts and related data will be transmitted off-host. In an image-generation context, prompts may contain sensitive business, personal, or proprietary information, so lack of explicit disclosure creates a meaningful privacy and data-handling risk.
The documentation explicitly states that API keys are stored in a local file under the user's home directory, but it gives no warning about credential persistence, file permissions, or the sensitivity of that location. This can lead users to leave long-lived third-party API credentials on disk in plaintext or weakly protected form, increasing the risk of accidental disclosure through backups, shared accounts, or local compromise.
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
# Seedream (default)
node {baseDir}/../../uni-image-proxy.js &
curl -X POST http://127.0.0.1:18800/v1/images/generations \
-H "Content-Type: application/json" \
-d '{"model":"doubao-seedream-5-0-260128","prompt":"a cute cat","size":"1024x1024"}'
The title and several usage references rely on Chinese-language labels such as 统一多平台 AI 图片生成, 绘画助手, and 模型平台, while the document otherwise targets a broader audience in English. This can amount to a language/locale constraint without explicit opt-in or an offered language choice.