The Agent Skills Directory

[COMMAND_EXECUTION]: The provided code example in tests/utils/baseline.ts uses child_process.execSync to execute shell commands with unsanitized input.
Evidence: The getChangedBaselines function takes a baseBranch argument and interpolates it directly into a shell command: execSync(git diff --name-only ${baseBranch}...HEAD -- tests/screenshots/).
Concern: If an attacker can influence the branch name input (for example, through a malicious pull request title or a configuration value that the agent is instructed to use), they could execute arbitrary shell commands on the system running the tests.
Remediation: Avoid using shell execution with string interpolation. Use child_process.execFileSync or the array-based arguments of spawn to prevent shell interpretation of parameters, or implement strict validation (e.g., regex ^[a-zA-Z0-9._/-]+$) on the baseBranch variable before use.

Playwright Visual AI Testing