The Agent Skills Directory

[PROMPT_INJECTION]: The instructions do not contain any patterns intended to bypass safety filters, override system constraints, or extract sensitive internal prompts. The use of a brand strategist persona is strictly for task alignment.
[DATA_EXPOSURE]: No hardcoded credentials, sensitive file paths, or network exfiltration patterns were identified. The skill instructs the agent to read a local context file (.agents/brand-context.md), which is a standard practice for maintaining project-specific information.
[OBFUSCATION]: The content is provided in clear text with no evidence of Base64 encoding, zero-width characters, homoglyphs, or other techniques used to hide malicious intent.
[REMOTE_CODE_EXECUTION]: The skill does not define any external dependencies, package installations, or remote script downloads.
[COMMAND_EXECUTION]: There are no shell commands, subprocess calls, or dynamic execution patterns present in the skill definition.
[INDIRECT_PROMPT_INJECTION]: The skill processes external data (brand materials and URLs) provided by the user. While this creates an ingestion point for untrusted data, the skill lacks any tool-using capabilities or shell access that could be exploited. The analysis is limited to generating text-based reports.

brand-audit