The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes Docker commands to manage containers and run tests. It interpolates user arguments ($ARGS) and the container path ($CPATH) directly into shell strings executed via bash -lc, which is vulnerable to command injection if malicious directory names or arguments are used.
[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect injection through user-controlled data. 1. Ingestion points: User-provided arguments ($ARGS) and the current working directory path ($CPATH) derived from HOST_PWD in SKILL.md. 2. Boundary markers: None present to delimit or ignore embedded commands. 3. Capability inventory: Executes shell commands within the container using docker exec and bash -lc as seen in SKILL.md. 4. Sanitization: No escaping or validation is performed on inputs before shell interpolation.
[EXTERNAL_DOWNLOADS]: Fetches the official Playwright environment from Microsoft's container registry (mcr.microsoft.com).

docker-playwright