The Agent Skills Directory

[COMMAND_EXECUTION]: The skill requires the agent to run standard development commands including npm run build and node server.js to manage the dashboard application. These actions are restricted to a specific local directory (/home/user/dashboard) to ensure environmental stability.
[EXTERNAL_DOWNLOADS]: The skill manages project dependencies via npm install, which involves downloading packages from the public npm registry. This is a standard and expected procedure for the described development workflow.
[PROMPT_INJECTION]: The skill retrieves and processes data from the agent's internal Turso database, which represents a surface for indirect prompt injection.
Ingestion points: Data is queried from database tables to populate dashboard components as described in the "Discovery and planning" and "Empty-database handling" sections of SKILL.md.
Boundary markers: The instructions do not define specific markers or guardrails for isolating potentially untrusted database content within the agent's prompt context.
Capability inventory: The agent possesses the ability to execute shell commands for building and serving the application and performing database operations.
Sanitization: No explicit sanitization or validation of database content is mandated prior to processing or display.
[SAFE]: The skill implements operational best practices for sandboxed environments, such as using symlinks to isolate ephemeral build artifacts (node_modules, dist) from persistent workspace storage, which prevents filesystem pollution and performance issues.

create-dashboard