Explanation Traces: Structured Decision Query

Overview

This skill reads the current session's session-trace.json and presents routing decisions, agent selections, skill phase transitions, and gate verdicts as a human-readable timeline. Every answer comes from what the trace actually recorded at decision time -- never from post-hoc reconstruction or rationalization.

Key constraints baked into the workflow:

• Read-only: this skill never modifies trace files or any other file
• Answers must come from recorded trace data, not from memory or inference about what "probably happened"
• If no trace file exists, explain how to enable tracing rather than guessing at decisions
• When the user asks about a specific decision, filter to that decision -- do not dump the full log
• Timestamps and evidence fields are authoritative; do not paraphrase away precision

Instructions

Phase 1: LOCATE