The Agent Skills Directory

[SAFE]: The skill follows a standard, safe instructional pattern for generating diagrams based on user input. No malicious code or hidden instructions were found.
[DATA_EXFILTRATION]: There are no network requests or attempts to access sensitive local files such as credentials or SSH keys.
[PROMPT_INJECTION]: The instructions do not contain any directives to override the AI's safety guidelines or bypass system constraints.
[INDIRECT_PROMPT_INJECTION]: The skill ingests user input to generate diagrams and uses an Edit tool to write to files. However, the risk is mitigated by a mandatory human-in-the-loop checkpoint ('Present Diagram' step) requiring the user to approve the generated code before it is saved to the file system.
[COMMAND_EXECUTION]: The skill does not execute shell commands or system-level scripts.

diagram-generator