The Agent Skills Directory

[DATA_EXPOSURE]: The skill accesses sensitive environment files such as .env, .env.local, and .env.development to detect server ports and host configurations. Although it instructs the agent to limit inspection to specific keys like PORT and HOST, reading these files presents a risk of exposing other sensitive credentials.
[COMMAND_EXECUTION]: The skill automatically detects and executes development commands (e.g., npm run dev, vite, next dev) from package.json to start local application servers. It also performs file system modifications in its 'Fix Mode' to address QA findings.
[EXTERNAL_DOWNLOADS]: The documentation suggests provisioning browser environments by running npx playwright install chromium if a first-class browser is not available on the host.
[INDIRECT_PROMPT_INJECTION]: The skill ingests and processes untrusted data from web applications during the rendering and QA process.
Ingestion points: The skill reads console errors, network failure messages, and DOM/accessibility snapshots from the rendered web application (SKILL.md).
Boundary markers: No specific boundary markers or instructions to ignore embedded commands within browser content are defined.
Capability inventory: The skill possesses the ability to execute shell commands for server startup and perform file writes for code fixes (SKILL.md).
Sanitization: No sanitization or validation logic is applied to the content retrieved from the browser tools.

fieldtest