The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: Fetches markdown references and instruction files from the author's public GitHub repository (Melvynx/ai-maxxing-skills) using curl at runtime to supplement project configuration.- [COMMAND_EXECUTION]: Executes system-level commands to manage project infrastructure, including git for source control, gh for GitHub repository management, and the cf CLI for Cloudflare deployment operations.- [REMOTE_CODE_EXECUTION]: Utilizes npx to download and execute remote packages such as lumail, convex, and shadcn during the application setup process.- [DATA_EXFILTRATION]: Transmits user-provided subscriber data (emails and metadata) to an external service (lumail.io) via API calls as part of the application's core sign-up functionality.- [INDIRECT_PROMPT_INJECTION]: The skill acts as a scaffolding engine that integrates user-defined briefs and processes untrusted inputs (like email addresses) in the resulting application, creating a surface for indirect prompt injection.
Ingestion points: User-provided application descriptions and the sign-up inputs in the generated web application.
Boundary markers: No explicit delimiters or safety instructions are defined to separate user input from generated code logic.
Capability inventory: Extensive system access including shell execution, file system modifications, and production environment deployments.
Sanitization: No explicit sanitization or validation logic is specified for the inputs handled by the generated application code.

create-software