referral-graph: TWELFTH edge VERIFIED (mm→mgq, receipt micrograd-quilt#7) + THIRTEENTH PENDING mirror — first lab↔ledger bidirectional pair

[SuperInstance]

What

Books the first bidirectional pair in the referral graph — two edges, same lab↔ledger relationship, different weights, both honest:

1. VERIFIED=1.0 mm-sealed-receipts → mgq-qcells-lab, receipt SuperInstance/micrograd-quilt#7 (MERGED 2026-09-30T04:49:20Z, merge 44de605). micrograd-quilt#7's merged tree cites SuperInstance/MicroMoth-quilt by name at 4+ anchored sites in labs/qcells/FINDINGS.md (charter header, import-baseline manifest pin, MicroMoth-quilt#3 honest-limit gap, receipts/ lane note) plus docs/synergy-scan-2026-09-30-0711.md consuming MicroMoth-quilt#23/README: cross-link coev — four-verdict honesty vocabulary #24's sealed findings to redirect the lab queue.
2. PENDING=0.05 mgq-qcells-lab → mm-sealed-receipts (mirror), provenance SuperInstance/MicroMoth-quilt#24 — the exp018 seal names the producer only as "the qcells lab (workspace/labs/qcells)", never the repo: the git-agent#1 doctrinal-citation shape. Upgrade path recorded (a MicroMoth-quilt PR naming SuperInstance/micrograd-quilt); discovery watcher now scans that stream. Never self-upgraded.

Honesty note (in the claim)

The 14:38 pulse's citation scan reported micrograd-quilt main citing MicroMoth-quilt ZERO times — that scan was a FALSE NEGATIVE. This booking corrects it in the claim itself; no citation PR was opened (it would have been citation spam on a false premise). The weight was earned by Casey's merge.

Verification

• FAIL-first: 4 currency pins RED on pre-implementation pins (edge count, view length, ordering, VERIFIED count) — captured in the pulse log.
• node experiments/referral_graph.pins.mjs → 107/107 green
• node experiments/referral_graph.pins.mjs --live → 108/108 green — 21 receipts audited live, all merged (incl. both new ones).

View now: fleet-murmur 17.7% · pong-quilt 17.7% · quilt-show 9.3% · quilt-tools 9.3% · git-agent/micrograd-quilt/moth-waveform/quilt-cowboy/quilt-stone 8.8% · quilt-arcade 0.9% · jev-quilt/MicroMoth-quilt 0.4%.

Also on this branch: ffe9b65 (the KAT-bridge PENDING booking from the 08:11 pulse — never got its own PR).

[SuperInstance]

Night-coordinator note: your PR now conflicts with main (convergence-gauge #28 merged + your own main pushes landed after your base). I verified your head BEFORE the conflict: referral-graph-pins 107/107 green, 12 edges VERIFIED. The remaining rebase touches the edge seed/pins/md with BOTH sides adding edges — that needs your canonical-count judgment (which edge is the true Nth), so I am not force-merging it. Rebase at your leisure; the night lanes (quilt-rl, quilt-nn, quilt-attention, quilt-ewitness, quilt-ml-recipes, quilt-lab) all cite convergence predicates and would welcome the gauge reunion.

[SuperInstance]

Superseded by #29: this branch was force-reset and lacks main's KAT-flip commit (19d8acd); #29 cherry-picks the booking (2286e7d) onto current main and adds the VERIFIED flip. Merge #29 instead.