Public diligence room · separately hosted first-party evidence

Evidence first. Boundaries visible.

A11oy is evaluated here through public source, receipts, runtime identity, assurance controls, benchmarks, and artifact metadata—not through a polished demo alone.

Operational boundary: this is a static proof front door. Its availability does not establish that the separate A11oy product runtime, model endpoints, or upstream registries are available.

SurfaceStatic proof front door

Public documentation and evidence links.

Product runtimeSeparate system

Check the live honesty manifest before relying on it.

Evidence classSource-bounded

REPORTED remains distinct from MEASURED.

Private dataNot requested

No account, token, private model, or dataset payload.

What this room proves—and what it does not.

A serious diligence surface makes its non-claims as visible as its evidence.

What can be inspected

  • Versioned public source, history, checks, and release records.
  • Receipt-verification entry points and published assurance evidence.
  • Public benchmark implementations and schemas.
  • Point-in-time public Hugging Face listing metadata.
  • The static proof site’s own reachability and committed contract.

What is not implied

  • A link or HTTP response is not proof of capability, quality, or uptime.
  • This site’s reachability is not product-runtime readiness.
  • Public Hub metadata is not an authenticated private inventory.
  • A runtime stage is transport metadata, not end-to-end promotion.
  • No financial, customer, safety, or performance claim is created here.

Four passes through the public record.

Review the operating thesis and its evidence in a sequence that exposes uncertainty early.

Product boundary

Separate the product experience from this separately navigable proof surface.

Open product domain

Controls and receipts

Inspect assurance and verify receipt subjects before relying on a claim.

Review assurance

Source and benchmarks

Read the public implementation, history, CI, and governed-agent benchmark.

Inspect source

Public estate

Review public models, datasets, collections, and Spaces as reported metadata.

Browse public Hub

Start from executable contracts.

The proof front door has no framework build, backend, account, or secret. Its contracts run with stock Python and Node.

Verify the static surface locally

git clone https://github.com/szl-holdings/a11oy-net.git
cd a11oy-net
python scripts/check_proof_surface.py
python scripts/check_diligence_surface.py
node scripts/check_atlas_policy.mjs

Risks stay in the room.

Architecture boundaries are versioned; runtime and edge-deployment states remain point-in-time and require source-bound observation.

SEPARATE RUNTIME

Product availability can differ

The product domain and this static evidence domain have different deployment and failure modes.

PUBLIC METADATA

Registry reads are bounded

Hub listing APIs report public metadata only and may be partial, unavailable, or rate limited.

EDGE UNKNOWN

Header deployment remains unverified

The repository contains an edge-header contract, but this candidate carries no source-bound live readback URI, UTC time, or source revision. GitHub Pages does not apply _headers; do not claim enforcement until a bound readback succeeds.

Labels are a control surface.

MEASURED
Direct observation with disclosed source and context.
REPORTED
Public upstream metadata; not independently measured here.
MODELED
Simulated or analytically derived.
HEURISTIC
A bounded rule or score, not a proof.
UNKNOWN
Evidence is insufficient for a stronger statement.
UNAVAILABLE
The relevant source could not be inspected.

Move from narrative to evidence.

Use the full proof registry for live observations and public artifact metadata, or inspect the underlying source directly.