The Agent Skills Directory

[PROMPT_INJECTION]: The skill contains standard instructional language and prompt templates for debugging. There are no attempts to override system instructions, bypass safety filters, or extract system prompts.
[DATA_EXFILTRATION]: No sensitive file access (e.g., credentials, environment variables) or network operations to external domains were identified. The suggested diagnostic snippets expose engine-level metadata (like FPS and draw calls) to the browser's global scope (window), which is standard practice for local development and debugging.
[COMMAND_EXECUTION]: The workflow involves standard development commands such as npm run build and npm run preview. There are no instances of privilege escalation (e.g., sudo) or suspicious command execution.
[REMOTE_CODE_EXECUTION]: The skill does not perform remote script downloads or execution. It references standard library initialization (e.g., HavokPhysics) typical for Babylon.js projects.
[INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to analyze console errors and game files. While this represents a data ingestion surface, the risk is inherent to the debugging task and the skill does not exhibit patterns that would facilitate the execution of embedded instructions from those data sources.
[OBFUSCATION]: All content is provided in plain text. No Base64, zero-width characters, or hidden unicode patterns were found.

babylon-debug-profiler