The Agent Skills Directory

[SAFE]: The skill provides structured instructions for manually persisting canonical knowledge, rules, and decisions to an ai-memory instance via the memory_write_page tool.
[CREDENTIALS_UNSAFE]: The skill instructions explicitly mandate a security best practice: the agent is told to never store live secrets or credentials in the memory pages and must redact them using a placeholder if encountered.
[INDIRECT_PROMPT_INJECTION]: The skill facilitates the creation of durable content (specifically under the _rules/ prefix) that will be read by the agent in future sessions. While this creates a persistence surface for instructions that could influence future agent behavior, it is an inherent function of a memory-writing tool and includes explicit step-by-step confirmation and path-based scoping to maintain data integrity.

aim-write