The Agent Skills Directory

[SAFE]: No malicious patterns, obfuscation, or unauthorized network operations were detected. The skill logic is transparent and aligned with its stated purpose.
[DATA_EXPOSURE]: The skill includes functionality to read existing system prompts from file paths using the Read tool. This access is triggered by user input and is restricted to the specific task of prompt optimization.
[INDIRECT_PROMPT_INJECTION]: The skill accepts untrusted data (prompts provided by the user) for analysis. While this creates a surface for indirect prompt injection, the risk is mitigated by explicit instructions to preserve existing safety and format constraints.
Ingestion points: User input via $ARGUMENTS and file content via the Read tool (SKILL.md).
Boundary markers: The instructions do not define specific delimiters for the input prompt data.
Capability inventory: The skill uses the Read tool; it does not request network access or file writing permissions.
Sanitization: No explicit filtering of the input prompt content is performed.

system-prompt-clinic