Clean-room adoption result: 30 August 2026¶
An independent coding-agent exercise passed the task's two integration-time
targets against exact library revision
4ca3a7c4f154d5478358702b64053dda9d0824f8:
| Exercise | Measured time | Target | Result |
|---|---|---|---|
| First verified action, including a material-drift refusal | 3.191415 minutes | under 60 minutes | Pass |
| Fifth distinct action | 0.537187 minutes | under 30 minutes | Pass |
The clean project installed the exact remote Git revision on CPython 3.13.14.
Its final suite passed 6 tests in 0.17 seconds and pip check reported no
broken requirements. The five actions funded a tax reserve, locked a purchase
budget, issued a travel advance, wrote off a receivable, and released a term
deposit.
What the exercise proved¶
- The first adapter reached prepare, bound authority, guarded execution, authoritative verification, replay safety, and material-drift refusal.
- Actions two through five reused the same authorization, semantic identity, atomic ledger mutation, verification, and runtime infrastructure.
- No Threvo repository knowledge or human implementation help was needed.
- The preserved clean-room artifact and its SHA-256 manifest reproduced with 6 passing tests after the timed exercise.
Friction found¶
The default Python on the machine was 3.14, outside the declared 3.11–3.13 support window. Selecting Python 3.13 resolved installation. Raw Markdown also hides rendered snippet includes and generated API signatures. The public site renders both, while source-only adopters can inspect the installed public signatures as directed by the coding-agent skill.
What it did not prove¶
This was an AI clean-room task using MemoryActionStore and deliberately
ephemeral protection. It did not qualify production storage, real identity,
managed key custody, a durable scheduler, retention operations, or an external
financial rail. It is also not the independent-human evidence required by the
broader integration-surface methodology
or the separate external-adoption gate.
The result therefore closes the task-specific coding-agent timing exercise, not production readiness or external-market adoption.