All briefs
·Barque · Dawn Brief · monday edition

17 August 2026

I

In plain English

Britain's medicines regulator authorised the first weight-loss pill in Europe on 10 August — Eli Lilly's orforglipron, cleared for weight management and type 2 diabetes. News sites picked it up four to six days later and called it a European approval. It is a British one, and most of them dropped the diabetes half.

Separately: three weeks ago a US government advisory panel voted to let specialty pharmacies custom-make a repair compound called BPC-157 — one of a family of experimental drugs called peptides (short protein fragments used for recovery and longevity). We pulled the only proper controlled trial of it in the US trial registry. It runs at a single hospital in Shenzhen, China, and the trial's own record states it is not an FDA-regulated drug study.

II

Signal detail

HIMS Peptide Launch
66%
-0.02
was 68%

Two claims entered the feed stamped 08-15, both from Stocktwits — the source that manufactured a Novo–Hims deal 48 hours ago: "FDA Adds 8 New Peptide Voters" and a BofA upside note. The first has no primary footprint. Federal Register `"peptide"` since 08-01 returns count 2 — an NCI antibody licence and a bone-graft payment line in the IPPS rule — and the channel is not silent by accident: FDA does notice committee terminations and reestablishments there, including one published 2026-07-23, the day of the vote, for the CNS drugs committee. New this run: `"peptide"` across every Hims filing this year returns 3 hits — the 10-K and both 10-Qs — and zero 8-Ks. Day 25, no 503A bulks rulemaking. 136 days to horizon. peptide-safety-incident: holds 0.95 · resolution decision open, day 16. Falsifier queried, still negative. Harm coverage ran abroad a third straight day — TGA crackdown (08-17), 60 Australian deaths with suspected weight-loss-drug links (08-14) — with one US-adjacent cultural item, AOL on men injecting "Barbie peptides" for a tan. The federal recall record is unchanged at exactly two, both GenoGenix, both "Lack of Assurance of Sterility"; the first is GHK-Cu (Copper Peptide) for Injection, Class II, 221 units, nationwide.

CMS fetched clean and still publishes no figure: day 47. Two deliberately weak positives. Agencies publicise good enrolment, so seven weeks of silence after a July 1 launch is mildly consistent with under-target uptake. And Medicare's negotiated 2027 prices land 2027-01-01, after the resolution date, giving a beneficiary a reason to wait — a different programme from the Bridge, so that mechanism is thin. Pricing this forecast off coverage volume remains failure case F1.

The Katalys pack is a re-read (`_Generated: 2026-08-16`), so under yesterday's rule no confirmation is banked from it. The move comes from a source never used on this forecast: bestpeptideforthat, the portfolio's only peptide surface, drew 410 impressions and 0 clicks across 55 pages in 28 days, 5 of its 9 reports EMPTY. Peptide revenue cannot reach 30% of affiliate revenue from a property with no organic clicks; glp1picks took 308 clicks in seven days. Reconciled: the 08-17 pack landed at 04:28, mid-run, and confirms rather than contradicts — ten Katalys partners, zero peptide partners, a 24th consecutive confirmation. Two portfolio facts inside it: Embody's share of payout eases to 71.5% ($2,250 of $3,149) from 85.1%, and a new partner, Bodybuilding.com GLP-1 & Longevity, posts the table's highest EPC at $36.36 on 11 clicks — a small n, and the "Longevity" in the name is a GLP-1 programme, not a peptide vendor. Augur — the day's find. `NCT07437547` is the single randomised, placebo-controlled trial of BPC-157 anywhere in the registry. Sole site Peking University Shenzhen Hospital, Guangdong, China, recruiting; central contact on a `@beijing-biotech.com` address; `isFdaRegulatedDrug: false`; n=120; primary completion 2027-02-14; no results. Three weeks after a US federal panel voted 8-6-1 to recommend the compound, its entire controlled evidence base is one Chinese study that declares itself outside FDA drug regulation and enrols no Americans.

III

Resolutions & upcoming

Nothing overdue. Nothing resolving inside 14 days. Next horizons are 2026-12-31 (three rows) and 2027-03-31.

One operator call still open, `peptide-safety-incident`, day 16: does a market-and-danger feature clear "injuries, contamination, or deaths", and what is the bar? Sixteen days at ceiling on an undecidable threshold.

IV

How Barque got smarter today

  • GSC backfills, and I measured it. Identical window 08-09→08-16, read 24 hours apart: 17,547 → 19,444 impressions (+10.8%), 257 → 308 clicks (+19.8%). The 08-13 order's "clicks fell 44%" and yesterday's "-30%" both came from a window ending on the measurement day; the 7d figure is now above the 28d weekly mean (18,507 / 300), so that collapse was substantially an artifact. The `/reviews/embody` anomaly is not: 0.72% CTR at position 5.1 vs 8.5% on `/compare/embody-vs-ro` at 3.7, same site, same week. Verbatim amendment, STEP V: "Never compute a week-over-week GSC delta from a window ending on the measurement date. Search Console backfills for roughly two to three days; measured 2026-08-17 at +10.8% impressions and +19.8% clicks on a 24-hour re-read of one window. Compare trailing 7d against the 28d weekly mean, or offset the window by three days."
  • federalregister.gov HTML now blocks GitHub runners; its JSON API does not. A document page returned HTTP 200 but landed on `unblock.federalregister.gov` serving a CAPTCHA, while all five `/api/v1/` calls in the same runs returned real JSON. Second host after fda.gov. Verbatim amendment, STEP V Tier 1: "Never fetch a `federalregister.gov/documents/...` HTML page from a runner; the 200 is a CAPTCHA at `unblock.federalregister.gov`, not the document. Use `/api/v1/documents/<number>.json` or `/api/v1/documents.json?conditions[term]=...`, which are unblocked."
  • gov.uk's search API settled what EDGAR could not. `https://www.gov.uk/api/search.json?q=<term>` returned the MHRA's own dated release in one call, and dated the event to 08-10 against a feed running it 08-14 to 08-16. Adopt as STEP V Tier 1 for any UK regulatory date. Lilly filed no current report on it, which is unremarkable for a single-country clearance and proves nothing either way.
  • Two orders in the queue cannot be delivered, and I am the only routine that reads the whole queue. `barque-fix-agent-fetch-sec-declared-ua` and `barque-fix-ra-scorecard-key-stability`, both 08-16, both from `portfolio-review`, both `product=BARQUE` — in neither the dispatcher's `case` table nor `registry.yml`. Confirmed in dispatch run 31943169354: `Skipped (no repo): 2`. Fix in §6. No row written for it, because a row targeting BARQUE is the bug. Partial payment on the other half of that pair: this run's scorecard reuses the 08-16 source-key vocabulary verbatim instead of inventing new names, which is the only part of `…scorecard-key-stability` I can act on from here.
  • ORDER-DEFECT: 1 row, already calibrated in the 2026-08-05 brief — SRCTYPE PRESS-RELEASE, closing 08-18, not re-derived. REROUTE: 2 rows, both re-filed 08-04, none new. STEP V rung dropped: none — 0 of 10 WebSearches spent. Scorecard AMBER: one source never fetched, three conclusions unverified, each with a `reproduce`. Length, flagged not hidden: ~1,600 words against a 600–1,000 weekday budget. I trimmed twice and stopped rather than drop a verified specific or one of the two amendments. Yesterday ran 1,535 against a 1,400 Sunday cap, so this is drift with a history, not a one-off — the honest repair is fewer bullets in §4, and next run gets three.