The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the Bash tool to execute shell commands for managing the impeccable CLI (e.g., npx impeccable install, npx impeccable update) and Maestro-specific lifecycle operations (e.g., maestro run complete).
[EXTERNAL_DOWNLOADS]: The skill facilitates the installation of the impeccable tool from the npm registry using the npx command when local resolution fails or a manual update is requested.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from project files and user instructions, creating a vulnerability surface for indirect prompt injection. \n
Ingestion points: The skill reads project-level context from PRODUCT.md, existing source code files, and user-supplied UI requirements. \n
Boundary markers: The skill employs a structured communication protocol using Maestro-specific symbols (→, ✓, ◆) and requires the creation of a formal report.md artifact to maintain task state. \n
Capability inventory: The skill is granted access to high-privilege tools including Bash, Write, Edit, Read, and spawn_agent. \n
Sanitization: Security is enforced through defined verification steps (e.g., confirmation pass 1/1) and mandatory human-in-the-loop confirmation gates (◆ GATE) before finalizing design changes.
[DYNAMIC_EXECUTION]: The skill dynamically routes execution and loads specific markdown reference files (e.g., reference/routing.md, reference/craft-floor.md, ui-codify.md) from the local file system based on user-supplied arguments and detected project state.

maestro-impeccable