Security audit
Security checks for vulnerabilities and agentic risk
This is a music-generation skill with no bundled executable payload, but its activation scope and declared tool authority are broader and less consistent than its stated purpose.
Review this skill before installing because it may be selected for tasks outside music generation. Use it only for ACE music/song generation, provide API keys through environment variables, and be aware that the referenced generate.sh script is not included in the package.
The trigger description is overly broad and unrelated to the skill's actual scope, encouraging activation for generic AI and orchestration tasks instead of music generation. In an agent platform, bad prompting/selection boundaries are security-relevant because they can route sensitive or unrelated workflows into a skill that exposes exec, read, and write capabilities.
The manifest and top-level description broaden a narrowly branded music-generation skill into generic AI model calling, intelligent dialogue, agent orchestration, and LLM application use. This scope inflation can cause agents to invoke the skill in unrelated contexts and grant exec/write-capable behavior outside the user's intended task, increasing the chance of misuse or unsafe execution.
The description explicitly states '支持中文交互' as part of the skill behavior, and the document is written as if Chinese interaction is the default mode without an explicit opt-in or language selection mechanism. This can violate language/locale policy when users have not chosen Chinese as their preferred interaction language.
The documentation advertises generic create/query/modify/delete and import/export/configuration operations that do not correspond to the stated music-generation use case. In a tool-using agent environment, such ambiguous capability claims can mislead orchestration logic or operators into treating the skill as a broader automation primitive, which is risky given exec/write access.
The activation condition is vague and spans broad media tasks beyond the documented capability set. This increases the chance of accidental invocation and unsafe command execution under false assumptions about what the skill can safely handle.
The trigger conditions expand the skill to video processing, audio editing, media conversion, and dubbing generation, which exceeds the documented purpose and can lead to inappropriate activation. Because the skill is allowed to use exec/write, overbroad triggers raise the risk of executing shell-based workflows in contexts where the user did not intend this tool to run.
These comparison claims present built-in error recovery and security checks as implemented features. In this file, there is only descriptive documentation and command examples; no concrete code or procedures are shown that would substantiate those built-in protections or recovery mechanisms.
No suspicious patterns detected.
