The Agent Skills Directory

[COMMAND_EXECUTION]: To facilitate user review, the skill uses shell commands (open on macOS and xdg-open on Linux) to display the generated HTML blueprints in the system's default browser.
[EXTERNAL_DOWNLOADS]: The visual blueprints reference the Mermaid.js library and the ELK layout engine from cdn.jsdelivr.net, as well as typography from fonts.googleapis.com. These are standard, well-known services for web development and documentation tools.
[PROMPT_INJECTION]: The skill's exploration workflow involves subagents scanning the codebase to synthesize implementation plans. As untrusted data from the codebase enters the agent context (SKILL.md, Step 1) without explicit boundary markers or sanitization, there is a surface for indirect prompt injection. Malicious instructions embedded in source code could potentially influence the generated task list or the verification commands provided in the final checklist. Users are advised to review all generated shell commands before execution.

blueprint