siftlisted

# Sift Do not vendor an outside system because it looks useful. Inspect only enough to reach a verdict. Evaluate: 1. **Need:** what current observed failure or missing capability would adoption solve? 2. **Redundancy:** does Loop-Dee-Loup, the target repository, or another authoritative shared repo already provide it? 3. **Provenance:** what license, attribution, redistribution, security, and supply-chain obligations apply? 4. **Fit:** does it preserve disposable sessions, vertical slices, compressed state, and target-repository authority? 5. **Cost:** is copying one mechanism cheaper and safer than adopting the whole system? 6. **Verification:** what deterministic evidence would prove the adapted mechanism works? Return one verdict: ADOPT, ADAPT, REIMPLEMENT, DEFER, or REJECT. Name the smallest useful piece, required obligations, destination authority, validation method, and what must not be imported. Do not create backlog work from speculative value. Only a founder-accepted, independently valuable outcome may become backlog work — a priority-horizon-labeled GitHub Issue, or, in a target repository, that repository's own Burn Order.