The Agent Skills Directory

[SAFE]: No malicious patterns, obfuscation, or unauthorized data access were detected. The skill is designed for legitimate software development workflows.
[EXTERNAL_DOWNLOADS]: The skill integrates with Playwright for testing and quality assurance. Playwright is a widely-used, well-known automation framework, and its inclusion for testing purposes is a standard development practice.
[PROMPT_INJECTION]: The skill exhibits a potential surface for indirect prompt injection because it processes untrusted user data (feature descriptions) that could contain hidden instructions. (1) Ingestion points: The [feature-description] parameter in the /dev:implement trigger defined in references/implement.md. (2) Boundary markers: None identified; the input is interpolated directly into the workflow context. (3) Capability inventory: The skill possesses capabilities to write and edit files (Write/Edit/MultiEdit), launch subagents for parallel tasks (Task), and perform browser-based testing (Playwright MCP). (4) Sanitization: No specific sanitization or filtering of the user input is described.
[COMMAND_EXECUTION]: The skill utilizes tools for file manipulation and subagent task execution to implement software features. These capabilities are consistent with the skill's primary purpose of feature implementation and development orchestration.

feature-implementation