The Agent Skills Directory

[PROMPT_INJECTION]: Indirect Prompt Injection vulnerability surface.
Ingestion points: The skill reads and processes user-provided LEAF documents through the leaf review command and direct content analysis specified in SKILL.md.
Boundary markers: Absent. The instructions do not define delimiters or provide specific warnings to the agent to ignore instructions embedded within the markdown files being cleaned.
Capability inventory: The agent has the authority to execute shell commands (git and leaf CLI), modify the contents of files, and perform directory manipulations such as moving and renaming folders in SKILL.md.
Sanitization: Content from external documents is not escaped, sanitized, or validated before it is processed or rewritten by the agent.
[COMMAND_EXECUTION]: Local shell command and file system interaction.
Evidence: The skill executes several CLI tools including git status, leaf doctor, leaf review, and leaf checkpoint to manage the project workspace as seen in SKILL.md.
Evidence: The 'Migrate' section in SKILL.md includes logic for structural changes to the project directory, such as renaming seeds to 01-sprouts and moving item folders one by one based on findings from the leaf doctor tool.

leaf-clean