Security audit
Security checks for vulnerabilities and agentic risk
This skill coherently connects OpenClaw to Soundside’s remote media tools, with expected cloud processing and library actions but no evidence of hidden or malicious behavior.
Install only if you are comfortable sending prompts, media, and related project data to Soundside and its provider backends. Avoid secrets or regulated data unless approved, and require explicit confirmation before delete, update, paid generation, or email-sharing actions.
The skill routes prompts, media, and related metadata to an external MCP endpoint and explicitly brokers requests across multiple third-party AI providers, but it gives users no privacy or data-handling warning. This can cause users to unknowingly transmit sensitive text, images, audio, or business information outside their local environment, increasing confidentiality and compliance risk.
The skill advertises lib_manage create/update/delete capabilities and lib_share email sharing without warning that these actions may modify, delete, or disclose stored resources. Users may invoke these operations through an agent without realizing they are destructive or externally sharing, leading to data loss or unintended disclosure.
No suspicious patterns detected.
