The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to perform multiple filesystem operations, including searching for files, reading content, creating directories, and renaming files (archiving) across global configuration paths (e.g., ~/.gemini/config) and workspace roots.
[INDIRECT_PROMPT_INJECTION]: The migration process involves reading legacy markdown files and interpolating their content into new SKILL.md files, which presents a surface for indirect prompt injection if the source files contain malicious instructions.
Ingestion points: Legacy .md workflow files located in user configuration directories and workspace roots.
Boundary markers: The instructions do not specify the use of delimiters or "ignore" warnings when processing the source content.
Capability inventory: Filesystem read/write, directory creation, and file renaming capabilities.
Sanitization: No sanitization, escaping, or validation of the source markdown content is performed before it is written to the new skill files.

migrate-workflows