The Agent Skills Directory

[SAFE]: The skill is entirely informational, containing markdown documentation files. It does not include executable code, automated setup scripts, or hidden payloads.\n- [PROMPT_INJECTION]: The skill includes behavior-shaping instructions for AI models (e.g., 'CRITICAL GUIDELINES FOR AI MODELS' in api_sdk.md) designed to optimize framework performance and ensure reliability. These are legitimate educational components of a technical reference skill.\n- [EXTERNAL_DOWNLOADS]: Contains documentation on installing official framework SDKs and utilities from standard, well-known package repositories (npm/PyPI) and references official framework repositories on GitHub.\n- [REMOTE_CODE_EXECUTION]: Describes the architecture and use of a sandboxed code execution tool ('run_code') as a core feature of the framework for handling programmatic tasks, which is standard functionality for the described service.

letta