The Agent Skills Directory

[COMMAND_EXECUTION]: The skill's primary workflow involves the agent generating Javascript scripts in the /tmp/ directory and executing them using node or npx. This dynamic execution allows the agent to perform complex browser-based tasks based on instructions. \n- [EXTERNAL_DOWNLOADS]: The skill uses npx playwright test, which may trigger the download of the Playwright framework and browser binaries (Chromium, Firefox, WebKit) from official registries. \n- [PROMPT_INJECTION]: The skill includes patterns for web scraping and content extraction (page.content(), extractTable()), which creates an indirect prompt injection surface. Malicious instructions on a target website could potentially influence the agent's subsequent logic. \n
Ingestion points: page.content(), allTextContents(), and extractTable in SKILL.md. \n
Boundary markers: None present in the provided templates to isolate external content. \n
Capability inventory: Shell access via Bash tool, script execution via node, and file system write access. \n
Sanitization: No sanitization or validation of extracted web content is included in the examples.

playwright-automation