The Agent Skills Directory

[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface during its primary generation workflow. * Ingestion points: The agent ingests untrusted data from the user prompt interpolated into server commands in SKILL.md, as well as external image URLs processed via WebFetch and theme.html files. * Boundary markers: Explicit delimiters or instructions to ignore embedded commands are absent in the prompt interpolation. * Capability inventory: The skill has access to the Bash tool, file system manipulation (Read, Write, Edit), and deployment capabilities to Cloudflare. * Sanitization: No specific sanitization or filtering logic for user-provided prompts or metadata blocks is defined in the instructions.
[COMMAND_EXECUTION]: The skill executes multiple local scripts using the bun runtime to manage the development lifecycle. * Authenticates the CLI and manages tokens using scripts/lib/cli-auth.js. * Launches a local editor and preview server via scripts/server.ts. * Assembles the final application into an HTML template using scripts/assemble.js. * Performs edge deployment to Cloudflare using scripts/deploy-cloudflare.js.
[EXTERNAL_DOWNLOADS]: The application template fetches core framework and styling libraries from well-known and trusted CDNs. * Loads the Babel standalone compiler from unpkg.com. * Fetches the Tailwind CSS browser runtime from jsdelivr.net. * Imports React, multiformats, and Fireproof modules from esm.sh.

vibes