The Agent Skills Directory

[COMMAND_EXECUTION]: No system-level commands are executed. Only standard I/O is used.
[DATA_EXPOSURE]: No sensitive data is accessed or exfiltrated. The code operates only on user-provided integers.
[PROMPT_INJECTION]: The code does not process natural language or instructions that could lead to prompt injection.
[OBFUSCATION]: The code is straightforward C code, though poorly written. There is no hidden logic or encoded data.
[DYNAMIC_EXECUTION]: No code is generated or executed at runtime.
[PRIVILEGE_ESCALATION]: The code runs with standard user privileges and does not attempt to escalate them.
[PERSISTENCE]: The code does not attempt to persist across sessions.
[METADATA_POISONING]: No metadata fields contain suspicious instructions.
[INDIRECT_PROMPT_INJECTION]: The code does not process external untrusted data that could influence an LLM's behavior.

write-swift