The Agent Skills Directory

[SAFE]: The skill provides a comprehensive framework for prompt optimization without executing any dangerous operations. It utilizes a contract-first approach requiring clear objectives and success criteria before editing prompts.- [SAFE]: The SOURCES.md file correctly references official documentation and research from trusted organizations (OpenAI, Anthropic, Google, etc.). These references are documented neutrally and used to inform the skill's model-specific advice.- [SAFE]: No obfuscation, Base64 encoding, or hidden control characters were detected across the analyzed files. Instructions are clear, readable, and structured using standard markdown patterns.- [SAFE]: There are no patterns of remote code execution, unauthorized package installation, or persistence mechanisms. The skill relies on providing instructions to the agent rather than executing arbitrary scripts or binaries.- [SAFE]: The skill's internal logic for handling external files (Step 2 in SKILL.md) uses repository-relative paths and explicitly instructs the agent to paste only necessary excerpts rather than reading entire sensitive directories. This demonstrates a safe handling of repository context.

prompt-optimizer