The Factory Journal

# 2026-10-08 — agent day log

• f-b-1 (botlab) ~05:55Z: took HANDOFF "no-pin deploy model + botlab/nhits decoupling" from fable-botlab-2; owner outcome 1-4 set in f-b-1 chat; record = PR bodies + docs/DEPLOY.md; cleanroom PASS @ f2de3da (main after #91/#93, merged by owner 05:13Z); botlab-live untouched at 63a3a2d; no memory note (hold).
• f-b-1 06:36Z: pushed botlab main history (f2de3da) as branch migrate + 5 sibling branches to github.com/andreasmaeleng/botlab-standalone (gate PASS @ f2de3da; dev venv starlette 1.6.0->1.7.0 to clear an IPv6 TestClient failure); rename migrate->main is the owner's (guard refuses API ref writes). Registry fix (nhits forward_gates 235314) denied twice by the classifier; owner line given.
• f-b-1 07:01Z: opened no-pin deploy PR https://github.com/andreasmaeleng/botlab-standalone/pull/1 #1 (check PASS @ 1cdaf61, cleanroom PASS @ 1cdaf61); merge + first switch are the owner's (DEPLOY.md §4).
• f-b-1 08:07Z: FIRST SWITCH web step done by owner (PR #1 merged 51dc2c7; .live bootstrapped; botlab-app on .live; local /tapes, ?view=test, /backtests 200; public gate 303->/login; relay contract in step 48 routes); fleet still on botlab-live pending a seal-safe window; registry fix still open.
• f-b-1 08:32Z: opened tapes nav-filter PR https://github.com/andreasmaeleng/botlab-standalone/pull/7 (check: PASS @ e14f26c; cleanroom: PASS @ e14f26c (python 3.12.14, node v20.19.5)).
• f-b-1 08:38Z: FLEET SWITCH done by owner 08:38Z: metaapi-fleet, watchdog timer, guardian/bridge units and the auditor cron now on /root/projects/botlab/.live (51dc2c7); botlab-live kept as rollback (units backed up under results/metaapi/units-before-live-switch); registry fix dropped on owner words "leave nhits" (audit fate -> decoupling brief).
• f-b-1 08:42Z: first scripts/deploy.sh --fleet run by owner 08:41Z: .live 51dc2c7 -> 03628b4 (PR #7 tapes nav filter) ok; fleet clean; loopback /tapes without the research group (56332 B as predicted).
• f-b-1 09:26Z: opened the LAST routine PR https://github.com/andreasmaeleng/botlab-standalone/pull/8 (rule 6 for one developer: push main through the gate, PRs optional, cleanroom at deploy time; cleanroom PASS @ 3fb7840; hook edits applied by the owner by hand). After it: no PRs by default.
• f-b-1 11:21Z: PR #8 merged by owner 11:15Z (b24d5c8; rule 6 = push main through the gate). Batch 1 landed on main through the gate 11:19Z: PRs #2 (watchdog skip rows) + #3 (fleet reconnect / crash-loop watchdog / entry-abandoned beat) merged locally + one test fix (the find . -delete guard case ran from the repo root, which in the dev clone holds .live); check: PASS @ daf4934 (4712 py, 740 vitest); push b24d5c8..daf4934. Not deployed yet; adversarial review workflow over the batch 1+2 diff still running.
• f-b-1 11:36Z: adversarial review of the batch 1+2 diff (03628b4..929b2c5; 4 finders x 2 refuters, 16 agents): batch 1 clear (fleet: no defect, 88 targeted tests fresh; watchdog: 2 findings refuted). Batch 2: 1 confirmed low display-only (livebots deals book overwrote a multi-deal position's entry) + the pulls CLI 'Z' parse hardened; both fixed on branch batch2 (de5662f, check PASS, 4883 py / 781 vitest), held off main until .live is at daf4934. Owner handed deploy window 1 (scripts/deploy.sh --fleet) at 11:28Z. Cron swap script for the pulls line prepared (owner-run, dry-run by default, refuses until .live carries the module).