The Agent Skills Directory

[DATA_EXPOSURE]: The skill attempts to access a local file at .agents/brand-context.md to gain context about the user's business brand. This is a local file-read operation within the agent's workspace and does not involve sending data to external domains or exposing sensitive system credentials.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from both user input and the local .agents/brand-context.md file. While it lacks explicit boundary markers or sanitization for this content, the risk is negligible because the skill does not utilize any high-risk capabilities such as code execution, network requests, or file-writing tools that could be leveraged by an attacker.
[NO_CODE]: The skill consists entirely of Markdown-based instructions and metadata. It does not include any scripts (Python, JavaScript, Shell) or binary executables, significantly reducing its attack surface.

personal-brand