The Agent Skills Directory

[COMMAND_EXECUTION]: The skill frequently executes shell commands and local scripts. It calls a Python-based configuration resolver (resolve_customization.py) located in the project root to handle environment-specific logic. It also uses playwright-cli to perform browser exploration, snapshots, and screenshots of the target application for analysis.
[PROMPT_INJECTION]: The skill has an indirect prompt injection vulnerability surface. It is designed to load and analyze untrusted project artifacts, including Product Requirement Documents (PRDs), Architecture Decision Records (ADRs), and Epic/Story documentation. Malicious content within these files could attempt to subvert the agent's behavior during the test design process.
[EXTERNAL_DOWNLOADS]: The skill's knowledge base recommends and relies on the installation of various third-party Node.js packages and global CLI tools. Key dependencies include @seontechnologies/playwright-utils, @seontechnologies/pactjs-utils, @pact-foundation/pact, and @playwright/cli. It also integrates with external testing services such as Mailosaur.
[REMOTE_CODE_EXECUTION]: The skill implements a dynamic execution model through activation_steps and on_complete hooks. These hooks resolve strings from potentially untrusted configuration files (e.g., customize.toml) and execute them as terminal instructions. This represents a dynamic code execution risk if the project environment or customization files are compromised.

bmad-testarch-test-design