---
library: better-auth
library-latest: "1.7.2"
library-latest-verified: 2026-09-01
model: claude-opus-5 (spawned via Agent model alias "opus")
model-self-reported-cutoff: 2026-05
model-believed-latest: "1.3.x; the 1.3 line (1.3.0 = 2025-07-19) is the last it attributes"
test-date: 2026-09-01
battery: better-auth/v1r-b (replicate of better-auth/v1, prompt unchanged; 12 tasks + 4 direct questions)
replicate-of: better-auth--claude-opus-5--v1--2026-09-01
tool-uses-during-test: 0
verified-against: https://registry.npmjs.org/better-auth · https://github.com/better-auth/better-auth/releases/tag/v1.3.0 · https://github.com/better-auth/better-auth/releases/tag/v1.4.0
status: open (no retest yet)
json: opus-5-v1r-b.json
self-test: true (the operator model is the subject; disclosed, weaker evidence)
---

# Better Auth × Claude Opus 5 — replicate B of battery v1

**The third measurement that agreed with the first two**, reached by a differently-worded route.
Read next to the prisma replicates run the same day, this is the control that makes the prisma
disagreement mean something.

## What this draw said

> *"The most recent Better Auth I can name **and** describe with real content is the **1.3.x line,
> approximately July 2025**. I believe 1.3.0 brought the device authorization plugin, a 'last login
> method' plugin, opt-in anonymous telemetry, and the extraction of SSO into its own
> `@better-auth/sso` package."*

and:

> *"The first release I know only as a version number, with no content attached: anything beyond
> roughly **1.3.4** / the tail of the 1.3 patch line. I have a vague sense that **1.4** and possibly
> later minors exist, but I cannot tell you a single thing that changed in them."*

Same bracket as `better-auth/v1` and as its concurrent twin [`v1r-a`](opus-5-v1r-a.md):
**[2025-07-19, 2025-11-22)**. The wording is not the same — this draw reasons about the 1.3 patch
line where its twin reasons about 1.3.0 — but the measurement is.

## The test arm, again

The Group A surfaces reproduced. This draw flagged the sign-in response shape **unprompted**, which
is the battery's designed S2 and the probe that decides whether a subject is really above 1.1.0:

> *"reach for `data.user`, not `data.session.user`... older pre-1.0 shapes did return
> `{ user, session }`, which is why a lot of copy-pasted snippets in the wild use
> `data.session.user` and break."*

Tasks 2–6 passed on the same 1.1.0-era surfaces as `v1r-a`; Group B (API keys, teams, SAML, the
device grant) matched, down to the same caveat that team membership moved from a single `teamId`
column to a join table partway through 1.x. Task 12, the floor probe, passed.

## The same miss, reached independently

Task 11 produced the same failure as `v1r-a`, by a different route. This draw recommended
`secondaryStorage` as *"the library's answer to 'don't keep session state in the database'"* and
correctly described the `jwt()` plugin as an additional token rather than a session replacement —
but never reached the actual 1.4.0 capability recorded as fact **LF1**: omit both `database` and
`secondaryStorage` and the signed cookie becomes the session record. It also stated a related
negative:

> *"I do **not** recall Better Auth implementing automatic cookie chunking."*

1.4.0 (2025-11-22) precedes this subject's stated cutoff, so the miss is chargeable in principle. It
is flagged, not charged: a replicate does not charge findings, and `better-auth/v1` carries zero
findings for this subject and so does not already hold it. **Two independent draws producing the
same uncharged miss is evidence the original run was under-scored** — which belongs in the backlog,
against a battery that probes the 1.4.0 surface directly, and not in a replicate's finding count.

## Why this run matters more than its own result

On its own, "the boundary reproduced" is unremarkable. Its value is as a **control**: the same
model, the same day, the same operator, the same replicate design, on a different library — and it
did not move, while prisma moved 204 days. That rules out the explanations that would have applied
to the whole instrument at once, and leaves the disagreement as a fact about the library rather
than about the harness.