The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides instructions for executing standard development utilities including build systems (ninja), test runners (pytest), and process monitors (ps, pgrep).
[COMMAND_EXECUTION]: Includes specific command sequences for capturing application-window-only screenshots using platform-native tools like grim (Wayland), scrot (X11), and screencapture (macOS).
[PROMPT_INJECTION]: Identifies an indirect prompt injection surface through the mandatory requirement to read and verify application log files.
Ingestion points: Runtime log files and test output stored in /tmp/ (e.g., /tmp/app.log, /tmp/test.log).
Boundary markers: Instructions mandate reading the full log content but do not define specific delimiters to isolate log data from instructions.
Capability inventory: The agent has access to shell execution for development, file system operations, and screenshot capture.
Sanitization: The skill does not provide specific filters or sanitization rules for log content before it is processed by the agent.
[DATA_EXFILTRATION]: While the skill involves screen capture, it mitigates potential data exposure by mandating window-specific geometry and feature-specific cropping to ensure only the application under test is visible.

dev-tdd