The Agent Skills Directory

[SAFE]: The skill operates primarily through complex natural language instructions and logical frameworks. It does not perform unauthorized network operations, sensitive file access, or command execution.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted user input in the form of 'wishes' or 'goals'.
Ingestion points: User-provided text strings (e.g., "I want to make a million").
Boundary markers: None explicitly present for user input, though the instructions focus on analytical parsing.
Capability inventory: No subprocess, network, or file-write capabilities exist outside of standard memory state persistence.
Sanitization: None, but the risk is negligible as the input is treated as data for formula decomposition rather than executable instruction.
[SAFE]: All external references, such as the documentation link to 'github.com/iamzifei/aikey', are directed at the author's official resources on a well-known service, which is considered a safe vendor resource.
[SAFE]: The use of a dedicated memory path ('{config.paths.memory}/aikey-path/') is a standard persistence mechanism for maintaining state in multi-step agent interactions and does not constitute a security risk.
[SAFE]: No obfuscation, hidden commands, or privilege escalation patterns were detected in the skill's logic or metadata.

aikey-path