The Agent Skills Directory

[SAFE]: The skill implements a robust orchestration pattern for a conversion pipeline, including mandatory preflight checks to verify MCP (Model Context Protocol) connectivity and tool availability before performing operations.
[SAFE]: Input gating rules require explicit confirmation of the source folder and target tool, preventing accidental execution on unintended directories.
[SAFE]: State management is handled locally via a .ui-convert/ directory (e.g., progress.json), adhering to best practices by keeping the conversion state outside of the agent's chat context.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user-provided project directories, which constitutes an indirect prompt injection surface.
Ingestion points: Reads source files from the project directory, as well as project.json and index.json (SKILL.md).
Boundary markers: Absent; there are no explicit instructions to the agent to disregard natural language instructions that might be embedded in the code or metadata being converted.
Capability inventory: File system writes (logs, state, registry) and MCP write operations via coordinated writer skills (SKILL.md).
Sanitization: None detected; the skill assumes the structural integrity of the project files it scans.

ui-convert-coordinator