The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes content from untrusted local sources which creates a surface for indirect prompt injection.
Ingestion points: Reads signal data from verification.json, review.json, and issues.jsonl in the .workflow/ directory (SKILL.md).
Boundary markers: None identified; the skill does not use specific delimiters to separate untrusted signal data from its internal instructions.
Capability inventory: The skill utilizes Bash for command installation and Write for file creation (SKILL.md).
Sanitization: The skill does not mention sanitizing or validating signal content before drafting patches or performing actions.
[COMMAND_EXECUTION]: The skill uses the Bash tool to execute a local utility command: maestro overlay add to install generated JSON overlays (SKILL.md).

maestro-amend