The Agent Skills Directory

[COMMAND_EXECUTION]: The scripts scripts/pl.js and scripts/start.js utilize child_process.spawn with the shell option enabled on Windows systems.
Evidence: The runCommand function in scripts/pl.js sets shell: process.platform === 'win32'.
Evidence: Arguments from process.argv are passed directly to the spawned process without sanitization. If the agent incorporates untrusted user input into these arguments, it could lead to arbitrary command execution on Windows platforms.
[EXTERNAL_DOWNLOADS]: The skill ensures the installation of the Playwright Chromium browser binary during the startup process.
Evidence: The ensureChromiumInstalled function in scripts/start.js calls pnpm exec playwright install chromium if the browser is not already found.
Note: This fetches configuration and binaries from Playwright's official sources.
[INDIRECT_PROMPT_INJECTION]: The skill enables interaction with external web content, creating a vulnerability surface for indirect prompt injection.
Ingestion points: Data enters the agent's context through Playwright commands such as goto and snapshot in scripts/pl.js.
Boundary markers: Absent. The instructions do not specify delimiters or warnings for the agent when handling content from external websites.
Capability inventory: The skill can execute shell commands via scripts/pl.js, write to the local filesystem (session metadata and logs) in scripts/start.js, and access the network through the browser.
Sanitization: Absent. There is no evidence of filtering or escaping external content before it is processed by the agent.

playwright-stable-profile