Skip to content

Incident Ledger

The incident ledger tracks problems and near-misses that should influence future planning.

Incidents should identify what happened, severity, whether it was resolved, and what to recheck later.

Near-misses are useful only if they remain visible. This ledger prevents safety and workflow risks from disappearing after a task completes.

  • Generated lessons
  • Review history
  • Harness docs
  • Automation plan

Known incidents and near-misses include CRLF test issues, STATE_SYNC self-reference, timeout misunderstanding, context loss, and source repo safety risks.

Incidents should be retained even when resolved. The goal is not blame; the goal is future avoidance.

  • CRLF issues: normalize line endings when they are not behavior under test.
  • STATE_SYNC self-reference: generated output must not become its own source.
  • Timeout misunderstanding: synchronous proof routes are not production async architecture.
  • Context loss: board history must supplement chat memory.
  • Source repo safety risks: board automation must never write to foryoutune-dev.
  • Whether resolved incidents have tests or process checks.
  • Whether Phase 2 introduces new safety risks.
  • Whether automation still preserves source boundaries.

Use docs/history-templates/incident-record.template.md for individual incident records.