UK-based independent AI efficiency projectEvidence before claims · 2026

Token Optimisation

AI efficiency · measured
Discuss a diagnostic

Provider-Free Resume Test: What the Harness Can Show

SYNTHETIC · V20 SCOPE ONLY

Provider-Free Resume Test: What the Harness Can Show

A long-scroll visual companion to the article. It summarizes the article’s evidence boundary and method; it does not extend the underlying result.

At a glance

Provider-Free Resume Test: What the Harness Can ShowFour-step visual: FAKE ADAPTER; EXERCISE INVARIANTS; NO PROVIDER CALLS; PRODUCTION UNKNOWN. This infographic presents the already-public V20 fake-adapter scope only; later test-case candidates are not combined or published here.SYNTHETIC · V20 SCOPE ONLY1FAKE ADAPTER2EXERCISE INVARIANTS3NO PROVIDER CALLS4PRODUCTION UNKNOWNEvidence labels and limits travel with the visual.
A fake adapter can test state transitions, idempotency, and checkpoint handling without exercising live quota behavior.

Evidence in scope

  • V20: 2,000 synthetic resume events
  • V20: 0 provider calls/tokens
  • Production recovery results: UNKNOWN

Evidence boundary: The lane/event counts and six-case test report come from separate versions and are intentionally not combined.

How to use the diagram

Assert job identity, wait class, backoff, reconciliation, replay safety, and verifier boundaries in the isolated harness.

What this visual explains: A fake adapter can test state transitions, idempotency, and checkpoint handling without exercising live quota behavior.

What it does not establish

No live wait duration, production resume, provider usage, or accepted production job is evidenced.

← View article · Return to source article