Skill: TraceMem Overview and Mental Model

Purpose

This skill provides the foundational mental model for using TraceMem. It explains what TraceMem is, why an agent should use it, and how to think about its core primitives: Decision Envelopes and Decision Traces.

When to Use

• When you are an AI agent performing actions that affect the real world (reading private data, changing state, spending money).
• When you need to prove why you did something to a human or auditor.
• When you encounter instructions to "use TraceMem", "open a decision", or "check policy".
• When you need to persist memory or context across different execution sessions (TraceMem is your long-term memory).

When NOT to Use

• For scratchpad operations, intermediate chain-of-thought, or temporary variable storage that does not need audit.
• For purely internal data processing that has no external side effects and strictly public data.
• When low-latency (<10ms) direct database access is required (TraceMem adds governance overhead).

Core Rules

• TraceMem is the Courtroom: You are the agent/defendant. You must present evidence (reads), cite laws (policies), and record your verdict (outcome).
• All Actions Require a Decision: You cannot read or write governed data without first opening a Decision Envelope.