The Agent Skills Directory

[Prompt Injection] (SAFE): The instructions use standard role-playing techniques ("You are an expert conversion copywriter") without attempting to bypass safety filters or ignore prior instructions.
[Data Exposure & Exfiltration] (SAFE): No sensitive file paths, credentials, or network exfiltration patterns were detected. The skill reads from a local context file (.claude/product-marketing-context.md) which is standard for project-specific customization.
[Unverifiable Dependencies & Remote Code Execution] (SAFE): There are no scripts, package manager files (requirements.txt, package.json), or commands that download and execute code from the internet.
[Indirect Prompt Injection] (LOW):
Ingestion points: The skill reads from .claude/product-marketing-context.md.
Boundary markers: None explicitly defined in the prompt interpolation.
Capability inventory: No executable code, subprocess calls, or network operations are present in the skill.
Sanitization: None present.
Risk Assessment: While the skill reads external data, it has no dangerous capabilities to exploit, resulting in a LOW risk profile.
[Obfuscation] (SAFE): No Base64, zero-width characters, or other obfuscation techniques were found in any of the analyzed files.

copywriting