Ln Spike

Retire one uncertainty. Output is knowledge, not code — spike code is throwaway, never promoted directly (Beck, XP). One question per spike; if multiple unknowns exist, run multiple spikes.

Spike notes are temporary. Reconcile durable findings into memory/SPEC.md and memory/PLAN.md instead of inventing standalone investigation docs unless the user explicitly asks for one.

Input

The question and approach: $ARGUMENTS

Investigate

1. State the question and what would constitute an answer.
2. Time-box the investigation. Stop when answered or exhausted.
3. Write a verdict.

Spike Verdict