The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill downloads the Playwright framework and the Chromium browser binary during its initialization phase. It also accesses Google Fonts via CDN during the infographic generation process.
[REMOTE_CODE_EXECUTION]: The skill executes the npx getdesign command to fetch and run an external styling tool from the npm registry at runtime. This allows for the execution of remote code from a package that is not verified or pinned.
[COMMAND_EXECUTION]: The agent is instructed to perform several shell-based operations, including initializing a repository with npm init, executing local scripts with node, and managing version control via git init, git add, and git commit.
[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface where untrusted data could influence agent or system behavior.
Ingestion points: Project metadata such as the project name, tagline, and features are collected from the user or extracted from existing project files.
Boundary markers: None identified. User-provided strings are directly interpolated into HTML templates.
Capability inventory: The skill can write files to the local file system (including in /tmp/ and the project assets folder), execute shell commands, and initiate network requests through the headless browser used for rendering.
Sanitization: There is no evidence of HTML escaping or sanitization for the user-provided project information before it is rendered by Playwright (Chromium), which could lead to code execution within the browser context if malicious content is processed.

readme-generator