kirokulisted

# Kiroku *The record of what actually happened.* Kiroku (記録, Japanese: record/documentation) manages the evidence trail that makes autonomous work observable. It is the implementation layer for Principle 2 (Observable Autonomy). ## Why this exists Autonomy without evidence is not delegation — it is abdication. When an agent (human or AI) does work autonomously, the question is never "did it do something?" but "can someone who wasn't there reconstruct what happened, why, and whether to trust it?" A trail that answers this question earns the right to more autonomy. A trail that doesn't answer it — or doesn't exist — means autonomy must be constrained regardless of how competent the work was. This is the core dependency: ``` Evidence → Trust → Autonomy ``` More evidence → more trust → more autonomy earned. Less evidence → trust decays → autonomy must be constrained. Zero evidence → zero trust → autonomy must be revoked. ## What good evidence looks like ### Multi-resolution Not everyone has the same time budget. A 50-page transcript is unreadable to the person with 2 minutes. A 1-paragraph summary is insufficient for the person with an afternoon. Good evidence exists at multiple resolutions simultaneously: | Resolution | What it answers | Time budget | |---|---|---| | **Digested** | Where are we? What just happened? Should I be concerned? | 30 seconds | | **Indexed** | What decisions were made? What was the reasoning? What changed? | 5–15 minutes | | **Full** | What e