The Agent Skills Directory

[PROMPT_INJECTION]: The skill instructions were analyzed for attempts to override agent behavior or bypass safety filters. No malicious patterns such as 'Ignore previous instructions' or DAN-style injections were found. The content is strictly focused on marketing frameworks.
[DATA_EXFILTRATION]: No commands for accessing sensitive local file paths (e.g., .ssh, .aws) or network exfiltration patterns were detected. The skill is entirely text-based.
[REMOTE_CODE_EXECUTION]: There are no patterns involving the download and execution of remote scripts (e.g., curl|bash). The README mentions a installation command using 'npx', which is a standard procedure for adding skills via the CLI, but the skill itself contains no executable logic.
[OBFUSCATION]: The skill files were scanned for Base64 encoding, zero-width characters, and homoglyph attacks. All text is in plain Markdown and no hidden payloads were detected.
[INDIRECT_PROMPT_INJECTION]: While the skill is designed to process user-provided copy and website text, it lacks any dangerous capabilities (no file writing, no network operations, and no tool access requested in frontmatter). This significantly limits the impact of potential indirect injections in the processed text.
[DYNAMIC_CONTEXT_INJECTION]: No dynamic context injection patterns using the '!command' syntax were found in the SKILL.md file.

storybrand-messaging-framework