The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests user-supplied information regarding their skills, resources, and business activities, which constitutes a potential attack surface.
Ingestion points: Untrusted data enters the agent context through user responses to triggers such as /aikey-product and other product-related queries described in SKILL.md.
Boundary markers: The instructions mandate a strict output format and adherence to specific "criteria" (判据), although they do not define explicit delimiters to separate user data from system instructions.
Capability inventory: The skill's capabilities are restricted to reading its own internal reference documents (e.g., aikey/references/家族公约.md, 认知框架.md) and writing to a designated local memory path (config.paths.memory). It lacks network access and shell execution capabilities.
Sanitization: No specific sanitization or filtering logic for external user input is defined within the skill.
[SAFE]: The skill does not perform any remote code downloads, utilize hidden obfuscation techniques, or attempt to access sensitive system files like credentials or SSH keys. Its behavior is consistent with its stated purpose as a product distillation advisor.

aikey-product