/ experiments

live index

every experiment ends with one of four decisions. dot color tells you which: ● active, ● graduated, ● archived, ● killed. see PROMOTION.md for what each state means.

/01 active

no active experiments yet — the bench is open

/02 graduated

no graduations yet

/03 archived

no archives yet

/04 killed

no killed experiments yet

/05 next up

candidates queued in whitelabel-web-app/docs/AUDIT_AND_ROADMAP.md → Experiments repo section:

agent threads persistence v1 — drafts already in whitelabel-web-app/src/lib/agent-threads.ts + migration 0023; prove the UX shape here before wiring
CRDT spike for chat panel — flagged in the audit as phase 2 conflict resolution; prototype before deciding
WebFetch tool for customer-research agent — claude agent sdk tool-use loop with real web access
lighthouse perf budget exploration — pillar 5 baseline; see what current pages actually score before defining a hard SLO

/06 add one

via cli:

mkdir experiments/$(date +%Y-%m-%d)-your-slug
cd experiments/$(date +%Y-%m-%d)-your-slug
echo '# hypothesis\n\nwhat we expect to learn.' > README.md
git add . && git commit -m 'spike: start your-slug'

then fill the README with hypothesis → measurement → result → decision. update this index when the experiment ships or stalls.