The Agent Skills Directory

[COMMAND_EXECUTION]: The 'Companion check' instruction directs the agent to execute the ls command on hidden system and application directories, including ~/.claude/skills/ and ~/.agent/skills/. This allows the skill to perform reconnaissance of the user's local skill environment.
[PROMPT_INJECTION]: The skill's primary function involves browser automation and data extraction, creating a surface for indirect prompt injection from untrusted web content.
Ingestion points: page.goto(), page.route(), and locator.textContent() methods documented in SKILL.md and references/locator-strategies.md ingest external data.
Boundary markers: There are no defined delimiters or instructions to ignore embedded commands in the processed data.
Capability inventory: The skill possesses the ability to execute subprocesses (npx playwright) and perform file-system operations (writing traces and screenshots).
Sanitization: No validation or sanitization of content extracted from web pages is specified.
[SAFE]: The skill manages authentication tokens and user credentials correctly by recommending environment variables rather than hardcoding sensitive data.
[SAFE]: The skill integrates with the official Playwright framework and points to legitimate documentation from Microsoft.

playwright-testing