The Agent Skills Directory

[DYNAMIC_EXECUTION]: The skill dynamically generates TypeScript/JavaScript code for Playwright end-to-end tests (e.g., e2e/tmp-screenshots.e2e.ts) and executes them using the Playwright CLI. This runtime generation and execution of code is a core capability of the skill but introduces risks associated with executing dynamically created logic.
[INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied instructions through the $ARGUMENTS variable to define the actions performed in the generated tests. There is a lack of boundary markers or sanitization logic to ensure that malicious natural language instructions do not result in the generation of harmful code, such as code that attempts to read sensitive environment files or perform unauthorized network requests during the test run.
Ingestion points: User input via $ARGUMENTS (SKILL.md).
Boundary markers: None provided; the agent is instructed to follow instructions directly into code generation.
Capability inventory: File system writes (mkdir, rm), git operations (git restore, git diff), and arbitrary code execution via playwright-cli test.
Sanitization: None specified for the conversion of natural language to test code.
[COMMAND_EXECUTION]: The skill executes multiple shell commands including playwright-cli, git diff, git restore, mkdir, and curl. These are used to identify changed components, manage temporary test files, and verify the local environment (e.g., checking if Storybook is running on port 6006).

screenshots