The Agent Skills Directory

[SAFE]: The skill is composed entirely of natural language instructions and evaluation criteria. It contains no executable scripts, binaries, or logic that could interact with the host system or network.\n- [DATA_EXPOSURE]: The skill mentions reading a local context file at .agents/brand-context.md. This is a standard practice for retrieving project-specific background information and does not involve accessing sensitive system files or credentials.\n- [COMMAND_EXECUTION]: There are no command line instructions or shell scripts present in the skill files.\n- [REMOTE_CODE_EXECUTION]: The skill does not perform any remote data fetching or code execution.\n- [INDIRECT_PROMPT_INJECTION]: The skill has a data ingestion surface through reading .agents/brand-context.md as specified in SKILL.md. While no explicit boundary markers or sanitization steps are defined, the skill possesses no dangerous capabilities such as network operations, shell command execution, or file-writing tools, which effectively mitigates the risk of indirect instruction exploitation.

brand-positioning