MentorMind Varro V1
Public factory baseline

Varro V1 is live, reviewed, and now has KG-2 runtime-collect evidence.

This page is the Build 1 review surface for the Varro documentation and tutoring site. It shows what exists now, how KG-2 checked source-backed promotion readiness in Rust, and how cycle 11 collected against the deployed tutor runtime contract without claiming a live model runtime.

Achieved Public shell deployed

The existing isolated Fly app serves this static Varro V1 surface over valid TLS.

Measured Route, eval, citation QA, output index, KG readiness, runtime contract, runtime collect

The series now tracks route verification, tutor eval pass rate, wall-clock, cost, static citation coverage, accepted useful output, source-backed readiness, runtime-boundary readiness, and collect evidence.

Constrained No infra mutation

Autonomous work can refresh content only; app, DNS, TLS, and secrets stay supervised.

Not claimed Not a promotion claim

The accepted-output index stays at 0.75 because source-backed KG promotion remains at 0/18.

Tutor seed

First learner path: write and check a VSL system.

The initial content is intentionally small: one concrete tutorial, a frozen eval set, a visible citation-check surface that requires citations or refusal, and a runtime contract surface generated to the KG-2 limit. That keeps the tutor measurable before any live model path is added.

  • Varro is a governed query and control surface, not the owner of truth.
  • The command grammar is `ask`, `show`, `check`, `run`, and `create`.
  • Preview is the default posture; execution requires explicit authority.
  • Out-of-scope or source-missing questions must be refused without fabrication.
Open citation checks Open runtime contract Open runtime collect
Minimal checked VSL
system tutorial.hello {
  mission "A first VSL system."
  authority lane "operator:local"
  domain software
  maturity draft

  context root = "helios://local/tutorial/hello"
  compile workspace-runtime -> varro

  runtime execution {
    host governed
    commit-mode host-only
  }
}
Evidence ledger

What the reviewer can check today.

Open release plan

Delivery

Public routeverified by Rust route checker
Target appmentormind-varro-web
Rollbackpinned image digest in runbook
Latest cyclecycle 11 collects QA and gap evidence from the deployed runtime contract

Measurement

Tutor eval1.0 pass rate on frozen v1 spike
Citation QA15 frozen eval items and 4 refusals checked by Rust route verification
Accepted output0.75 from reviewed surfaces, tutor eval coverage, and live deploy
KG readiness7 of 7 external source hashes verified by Rust
Runtime contract5 answer stages and 4 refusal boundaries generated to KG-2 limit
Runtime collectpre-collect public route verified; no usage or live runtime evidence claimed
Kickoff cost0.6104224 USD lower bound
Route signalin control after public verification

Knowledge graph

RecordsKG-2 draft source-linked rebuild
Relationshipsnormalised records, not inline claims
Promotion0 of 18 records are source-backed
Active gapspromotion execution gate; live tutor runtime
Refinement signal

Accepted output is now separated from activity.

The useful-output score counts only review/QA-accepted, source-traced formation. It gives credit for the public surface, evaluator, citation surface, route evidence, review evidence, and SPC series, but gives no KG-promotion credit until records move through a source-backed gate.

  1. 1Surfaces

    8 of 8 accepted Build 1 surfaces are visible or evidenced.

  2. 2KG

    11 records are externally hash-eligible, but 0 of 18 are source-backed; draft status is preserved.

  3. 3Tutor

    Frozen v1 eval coverage is 15 of 15 with pass rate 1.0.

  4. 4Deploy

    The live public route passes Rust verification.

Next controlled step

Improve content depth without weakening the gates.

Use KG-2

Use the collect evidence to decide whether the next cycle should rebuild KG-2 or spike the runtime adapter.

Integrate the runtime

Move from static citation checks to a reviewed tutor answer surface with citations enforced in the interface.

Compare approaches

Record cost, wall-clock, interventions, reviewer-orientation coverage, and final artifact quality for the just-do-it and specification-refinement passes.