- start.sh: COPYBOT_ROLE=live — unarmed (missing LIVE_PRIVATE_KEY /
LIVE_FUNDER_ADDRESS / LIVE_CONFIRM) idles in geocheck, no clone, no
book, no orders; armed runs config.live.example.json + own state file
copybot_state.live.json, --live, poll mode (webhook stays on paper).
- confirm_live: accepts the phrase from LIVE_CONFIRM env — the USER
types it into flyctl secrets set (rule 0.7 human checkpoint, headless);
wrong value aborts, unset disarms at next boot.
- config.live.example.json: Set E + pinned floors (mirrors paper),
bankroll $50 @ 10% = $5 stakes, caps untouched (rule 0.6), leaked
Discord webhook URL removed (ROTATE it — was committed publicly).
- fly.live.toml: separate app wwf-copybot-live, arn, no inbound, no
checks, role env baked.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Chain-sweep backfill (784k conds ≈ ~1k calls, not 8 days of per-cond
queries) → shadow audit → consumer flips shallowest-to-deepest →
trust.py simplification last. Ground rules codify the 2026-07-08
alignment-audit discipline: overlay never rewrite, one flip per day,
never coupled with follow-set changes, calibration continuity notes.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The live path gates PER TRADE; reconcile_entries gated on POSITION size,
so 0xbadaf319's paired YES+NO arb clips ($6-37 each, all correctly
skipped below his $45 floor) accumulated into $40+ positions that read
as downtime misses while the bot was demonstrably UP. Now a miss needs
some SINGLE buy on the market to pass the same follow filter (market-
filtered /trades, full history); sub-floor accumulations log-and-skip
with a persisted verdict (state.no_miss_toks). Four phantom records
removed from the book; their hypothetical P&L never belonged in the
missed counterfactual.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
New 'book' field in the fills ledger: {bb, ba, spread, bid5c, ask5c}
from CLOB /book at fill time. The calibration experiment's weakest
assumption is the flat SLIP haircut; weeks of these snapshots give an
empirical fill model for the backtest and a depth gate before real
money sizes up (today's probe: a stake vs $77 of bids within 5c).
Best-effort, never blocks a copy.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
copybot[bot]