copy-editing
Pass
Audited by Gen Agent Trust Hub on Aug 17, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill implements a structured framework for linguistic and marketing analysis of text. No malicious patterns, obfuscation, or unauthorized access attempts were found.
- [DATA_EXPOSURE_AND_EXFILTRATION]: The skill attempts to read from
.claude/product-marketing-context.md. This is a local, project-specific configuration file used to provide relevant business context for the agent's tasks and does not involve exfiltration or access to sensitive system credentials. - [INDIRECT_PROMPT_INJECTION]: As a copy-editing tool, the skill naturally ingests untrusted user content (marketing copy). While it does not define strict boundary markers for this input, its functionality is limited to text analysis and rewriting, without invoking high-risk tools or executing commands based on that input.
- [REMOTE_CODE_EXECUTION]: All referenced files are local to the skill package. There are no external dependencies, package manager invocations (npm, pip), or remote script downloads.
Audit Metadata