The Agent Skills Directory

[DYNAMIC_EXECUTION]: The skill promotes the use of Marshal.load(Marshal.dump(object)) as a recommended idiom for deep copying objects.\n
Evidence: Found in SKILL.md under the <ruby_idioms>/<deep_copy> section.\n
Evidence: Found in references/creational-patterns.md within the Prototype pattern implementation (deep_copy method).\n
Security Concern: Marshal.load is unsafe when used on data that could be influenced by untrusted input, as it can be exploited to achieve arbitrary code execution. Providing this as a standard solution without security warnings increases the risk of RCE vulnerabilities in implementations derived from these templates.\n- [INDIRECT_PROMPT_INJECTION]: The skill facilitates the generation and execution of Ruby code based on user-provided design descriptions, creating a surface for indirect injection.\n
Ingestion points: User-provided descriptions of design problems and pattern requirements as described in SKILL.md ("Use when encountering...").\n
Boundary markers: Absent. The skill does not instruct the agent to use delimiters or ignore instructions embedded in the user's design specifications.\n
Capability inventory: The skill explicitly suggests executing the resulting Ruby files via the shell in the <running_examples> and <success_criteria> sections of SKILL.md.\n
Sanitization: Absent. There is no mention of sanitizing or validating the logic within the user's requested design before implementation.

design-patterns-ruby