Skip to main content

Verification

The evidence, tied to a named release candidate.

Every number here carries its release, check, environment, source, and limit. Nothing is presented as broader than what was actually run.

Release under test

Supervisor V1

Version
1.0.0
Examined archive SHA-256
950172FD3F5B1C0D911E13140BA610ECCBAEF857832629CEDE1F11DD35B3E284
Status
Release candidate — not yet the finalized customer artifact. The examined archive is not offered for download.

Evidence

Pass results, with their limits attached.

Each row states its environment and what passing does not prove.

Verification evidence for the examined release candidate
ItemCheckEnvironmentResultEvidence sourceDoes not prove
Unit suiteAll unit cases in the bundle's verification harnessWindows PowerShell 5.1 (independent local run)319 / 319 passed · exit 0Independent isolated execution of this exact release candidate on 2026-09-06.Commercial readiness, broad correctness, or fitness for any particular repository.
Example DryRuns-Init then -DryRun per fixture; pass requires exit 0 and no state.jsonIsolated local git fixtures (PowerShell-based audit harness)4 / 4 passedSanitized output from the recorded dry-run audit of the examined release candidate.Real Execute-mode behavior, or that any code produced by an agent is correct.

Method

How the DryRuns were executed

  1. Four isolated fixture repositories were created in a temporary sandbox: a first run, a web/node application with sibling worktrees, a Python service, and a hardened configuration.
  2. Each fixture ran the engine's -Init once and then -DryRun once against the fixture.
  3. A case passed only when init exited 0, dryrun exited 0, and no state.json was written.
  4. The sandbox was removed afterwards. The displayed terminal text comes from the sanitized recorded output.

Source-control behavior

Observe changes, never change history.

  • Repository changes are observed through a read-only Git chokepoint.
  • The wrapper refuses stage, commit, push, merge, rebase, reset, clean, checkout, stash, tag, fetch, and pull.
  • No automatic integration or deployment exists in the examined release.

Provenance

Where these numbers come from

The unit result comes from an independent local Windows PowerShell 5.1 execution of this exact release candidate on 2026-09-06, finishing with exit 0. The DryRun results come from a separate isolated local audit whose sanitized transcript is the source of the homepage demonstration. Both remain tied to the examined release candidate and to nothing else.

Limits

What the harness does not cover

  • DryRuns exercise planning and gate behavior, not real Execute-mode agent runs.
  • The unit suite tests the engine, not arbitrary third-party libraries or every repository shape.
  • No result here is an independent external usability test.
  • No result here is a correctness, security, or production-readiness certification.

Straight talk

PASS is a review outcome, not proof of correctness.

A green result here means the reviewed condition was met inside the tested environment. It does not mean the product is ready to buy — the release remains a release candidate and commercial paid launch is NO-GO while the launch gates are open.

This does not prove the product is commercially ready, broadly correct, secure, or production-ready for every repository. The final verdict about what ships stays with a human.

See the control flow with your own eyes.

Review the demo criteria and current scheduling status. Any future walkthrough will use authentic product behavior.

Book a Demo