Skill: TraceMem Notes and Context

Purpose

This skill teaches how to "show your work" by adding context and reasoning to the decision trace. This makes the agent's internal logic visible to auditors and humans.

When to Use

• Before making a significant decision (e.g., "I decided to approve this because X").
• When model confidence is relevant ("Confidence: 0.85").
• To summarize findings from multiple reads ("Found 3 matching records, selecting the most recent").

When NOT to Use

• Do not use as a general logging facility for debugging (use your standard logging for that).
• Do not dump raw massive JSON payloads (summarize them).

Core Rules

• Show Your Logic: If you make a choice, record why.
• No Secrets: Never put API keys, passwords, or highly sensitive PII in the context summary.
• Structured Data: Use the payload field for machine-readable context (e.g., scores, counts), and summary for human-readable text.