The Agent Skills Directory

[DATA_EXPOSURE]: The skill manages project-related knowledge within a dedicated local directory (.omw/wiki/). It does not access sensitive system paths, environment variables, or hardcoded credentials.
[COMMAND_EXECUTION]: The skill utilizes basic shell commands (specifically rm) for managing the lifecycle of wiki files. These operations are scoped to the wiki's storage directory and do not involve privilege escalation or persistence mechanisms.
[INDIRECT_PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection as it processes untrusted markdown content.
Ingestion points: Files located under .omw/wiki/*.md and user-provided content during the 'Add' operation.
Boundary markers: The skill uses YAML frontmatter for metadata, but there are no explicit instructions for the agent to ignore or delimit potentially malicious instructions within the markdown body of the wiki pages.
Capability inventory: The skill has access to Read, Write, Edit, Grep, Glob, and Bash (for removal) tools.
Sanitization: There is no mention of sanitizing or validating the content of the wiki pages beyond frontmatter consistency checks during the 'Lint' operation. As the skill is designed for internal project knowledge, this risk is considered typical for file-processing tasks.

wiki