Skip to evidence
Makemake View Companion

Public engineering register · 25 August 2026

Engineering evidence,
not promises.

Makemake publishes what is released, what is implemented behind a gate, and what remains in development. Every claim below includes a scope, a verification basis and a limitation.

Read the verification View evidence JSON

Maturity ledger

A clear line between
today and tomorrow.

“Released” describes a checked public surface. “Implemented · gated” describes reviewed code that still depends on provider, entitlement or deployment work. “In development” is not a present-day customer capability.

Released

Public Companion presentation

The nature-led website and Google identity surface are deployed. The public presentation marks unreleased products as in development.

Open the presentation
Implemented · gated

Identity and access controls

Google is the only enabled sign-in provider in this release. Phone and SMS sign-in are disabled. Paid-entitlement evaluation awaits an authoritative ADAR projection. This register does not assert the private operational pipeline's deployment state.

In development

Cross-device coordination

PC, Robot, TV, Drone, IoT, Mobile and Car are planned endpoints. Makemake does not claim these integrations or cross-device task execution are available today.

Historical reviewed baseline

A dated, self-reported
release check.

This dated baseline identifies the exact pre-cutover revision, environment and command. It remains historical evidence until the Firebase cutover receives its own committed revision and completed release check.

Source revision
1e775bbe756cb93bf0136526753657efa4b95e86
Verified
24 August 2026
Environment
Windows release workspace
Command
pnpm run check
Outcome
Passed

This command runs provider policy, secret scanning, hosting policy, TypeScript checks, automated tests and production builds. The source repository is private, so this record is not publicly reproducible. It is self-reported release evidence—not an independent certification or a real-world device benchmark.

14Web tests
11Core tests
31API tests
19Worker tests
75Total automated test cases passed in the dated Windows verification

Two worker cases are Windows-conditional. The 75-case result refers specifically to the dated Windows run, not every CI operating system.

Implemented controls

Specific mechanisms.
Bounded claims.

These statements describe reviewed implementation and tests. They do not turn planned device products or a private operations pipeline into released customer capabilities.

  1. 01

    Paid access fails closed

    A role alone cannot unlock Companion. The reviewed presentation gate also requires a signed entitlement start and an unexpired end time. No customer action API is represented as released.

    Implemented · tested
  2. 02

    Private operations have a separate owner boundary

    The configured owner identity must match, use a verified email and use Google as the token sign-in provider. The matching API hardening requires its own service release.

    Implemented · tested
  3. 03

    Commerce authority stays with ADAR

    Makemake does not invent a product, price or payment role. Configured checkout links are accepted only over HTTPS on adar.asia or its subdomains, and no live checkout offer is asserted here.

    Implemented · gated
  4. 04

    Operational AI cannot publish directly

    The reviewed source binds proposals and images to hashes, requires local approval and fresh commerce authorization, then records a durable receipt before acknowledgement.

    Deployment not asserted
  5. 05

    Credentials and work are bounded

    Operational secrets use authenticated envelopes. Signed worker requests bind method, path and body hash within a replay window. This is a code-level control claim, not a penetration-test result.

    Implemented · tested
  6. 06

    Web delivery is policy checked

    The hosting policy requires restrictive security headers, blocks framing and object execution, and excludes unsafe inline or evaluated application scripts.

    Released · checked

Method

How a claim earns
a place here.

1

State it narrowly

Name the mechanism and scope. Avoid comparative or universal safety language without evidence.

2

Tie it to a revision

Record the source revision, verification date, environment and exact command.

3

Show the boundary

Publish what the control denies, what remains gated, and what has not been measured.

4

Update or retire it

Future releases must refresh the evidence record rather than inherit an outdated claim.

Limits and next evidence

What we do
not claim.

Makemake has not published cross-device completion rates, latency or uptime measurements, independent penetration-test results, accessibility certification, customer adoption metrics, or comparative rankings. Phone and SMS joining are disabled; paid checkout and customer Companion actions are not presented as verified live capabilities.

The next evidence milestone is a versioned cross-device task protocol with predefined tasks, environment details, repeated runs, failures, raw results and independently reviewable methodology. Until then, device coordination remains a product direction—not proof of market leadership.

Read the machine-readable limitations