The Agent Skills Directory

[SAFE]: The skill consists of markdown instructions and reference guides for stylistic and linguistic tasks. It does not perform any programmatic actions or network operations.
[PROMPT_INJECTION]: The skill is designed to ingest data from an external file (brand-brief.md) to adapt its tone and vocabulary, which presents an indirect prompt injection surface. Ingestion points: The skill looks for .claude/brand-brief.md or brand-brief.md in the project root to obtain brand-specific instructions and glossary terms. Boundary markers: There are no explicit boundary markers or instructions to disregard potential commands within the brand brief file; the skill is explicitly told to let the brief override internal rules. Capability inventory: The skill has no defined capabilities or tools for file writing, network access, or command execution, which significantly mitigates the risk of exploitation. Sanitization: No sanitization or validation of the brand brief content is performed.

texten-de