The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes various system commands and CLI tools to manage the development lifecycle. Evidence: Uses git for repository management, including branch creation and commits; Utilizes gh (GitHub CLI) for creating and managing issues and pull requests; Executes npm and npx commands to install dependencies and run the Playwright testing suite; Uses curl within test scripts to verify the functionality of local endpoints.
[EXTERNAL_DOWNLOADS]: The skill downloads and installs well-known development and testing tools. Evidence: Installs the @playwright/test package and Chromium browser binaries via npx playwright install as part of its automated testing setup.
[INDIRECT_PROMPT_INJECTION]: The skill processes user-provided prompts to autonomously perform code modifications and external interactions, presenting a potential injection surface. Evidence: 1. Ingestion points: Processes a user-provided <ask> via the /lfg command in SKILL.md. 2. Boundary markers: Lacks explicit delimiters for the user prompt, but uses a structured 'Shape-lite' phase to define assumptions before execution. 3. Capability inventory: Possesses capabilities to write files, execute shell commands, and interact with GitHub repositories. 4. Sanitization: Implements a 'Stop classes' mechanism to pause for user confirmation before performing sensitive actions like credential entry or destructive operations.
[DYNAMIC_EXECUTION]: The skill generates and runs code dynamically to verify the generated proof-of-concept. Evidence: Creates playwright.config.ts and TypeScript test files from templates to perform end-to-end testing; Generates and executes shell scripts to validate CLI and API behavior.

lfg