The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill provides instructions for the agent to scrape content from a web page using a Playwright probe script to diagnose UX issues and generate code fixes. If the target web page contains malicious instructions designed to manipulate the agent, there is a risk of the agent executing unintended actions, such as writing malicious code during the Fix step. \n- Ingestion points: The ux-laws-probe.js script (referenced in references/probe.md) extracts document.body.innerText and various attribute values from the target website. \n- Boundary markers: The skill does not provide specific instructions to the agent to treat the scraped content as untrusted or to ignore instructions embedded within the page text. \n- Capability inventory: The skill specifies a Fix step in SKILL.md where the agent is expected to repair broken flows and make changes to shared components or tokens. \n- Sanitization: No sanitization or validation of the scraped data is mentioned before it is processed by the agent.\n- [COMMAND_EXECUTION]: The skill provides a shell script in references/probe.md that uses the Playwright CLI to automate browser actions like opening routes, resizing the viewport, and taking screenshots. These operations are core to the skill's diagnostic functionality.\n- [EXTERNAL_DOWNLOADS]: The diagnostic process involves running npx --yes @playwright/cli@latest, which downloads the Playwright package from the NPM registry. Playwright is a well-known browser automation tool maintained by Microsoft.\n- [DYNAMIC_EXECUTION]: The skill uses the Playwright CLI eval command to execute the provided ux-laws-probe.js script within the browser context to collect page metrics and element data.

enhance-ux-laws