Write AI Agent traffic generation specification

[sehkone]

Problem

The overview document references "AI Agent 트래픽 생성" as a separate document, but it does not exist yet. Phase 2 and Phase 3 planning cannot proceed without this specification.

Scope

• Define AI Agent traffic classes (DB queries, API calls, LLM proxy communication, file system access)
• Specify policy violation scenarios and expected Ground Truth output
• Specify runaway scenarios and expected Ground Truth output
• Define how AI Agent activities interact with the Scenario Definition schema
• Describe the expected bundle artifacts for AI Agent scenarios

Out of Scope

• Implementation of AI Agent traffic generation
• AI Actor behavior (covered in existing AI Actor Behavior Model document)
• AI control enforcement mode implementation

Done when

• The document defines traffic classes, policy violation cases, and runaway cases
• Expected bundle and Ground Truth outputs are specified with examples
• The specification is concrete enough to write Scenario Definition extensions and Validator checks

Dependencies

• Freeze v1 contracts across Scenario, Bundle, Ground Truth, and Validator #9 (v1 contract should be frozen so AI Agent extensions build on a stable base)