AgEvidenceInvestor Evidence Room
Technical Reference ↗GitHub ↗
Evidence · code

Code evidence

Documents describe architecture; code demonstrates it. Everything here is labelled by what it actually shows — implemented, demonstrated, partial or planned.

3 layersLabelled by evidence type
Why this matters
Telling a reviewer to go and inspect GitHub is not diligence support. This index says which artifact answers which question, and refuses to let a roadmap read as an implementation.
Sources

The two repositories

Rust core + Python SDKmeronrudy/athian-evidence-bazaar ↗

Pinned ref: fcd7cf96f724db7b24602a0778de55247e2da532. Every count cited in this room is read at that ref.

Rails Evidence Planemeronrudy/AgEvidence ↗

Domain model, evaluation engine and review workflow. Demonstrated, not production-proven.

By layer

Repositories, tests, fixtures, demos

Rust Core

Rust
  • RepositoryImplemented

    Canonicalization, hashing, signing, verification

  • TestsImplemented

    Fixture parity and corruption rejection

  • FixtureImplemented

    Byte-stable reference objects

  • Verification demoDemonstrated

    Offline CLI verification of an exported package

Layer page →

Python SDK

Python
  • RepositoryImplemented

    Five primitives, validation, serialization

  • QuickstartDemonstrated

    First valid object in minutes

  • Example integrationDemonstrated

    Synthetic portfolio-shaped adapter

  • FixtureImplemented

    Reproducible reference payloads

  • SchemaImplemented

    Canonical evidence model schema

Layer page →

Rails Evidence Plane

Rails
  • RepositoryDemonstrated

    Evidence plane application

  • Domain modelImplemented

    Program, requirement, evaluation, determination

  • DemoDemonstrated

    End-to-end synthetic walkthrough

  • TestsPartial

    Evaluation and review flows

  • Production roadmapProposed

    What is deliberately deferred

Layer page →
Diligence

Claims ↔ code

ClaimLayerCode / proof
Evidence is deterministicRust Core
  • Rust fixture #001Implemented
  • Verifier test suiteImplemented
  • Canonicalization specificationImplemented
Integration is lightweightPython SDK
  • Python quickstartDemonstrated
  • Synthetic DIT adapterDemonstrated
Evidence can be reusedPython SDK
  • Shared primitivesImplemented
  • Cross-company fixture setProposed
Evidence can be evaluatedRails Evidence Plane
  • Rails Evaluation modelDemonstrated
  • Requirement engineDemonstrated
Review is attributableRails Evidence Plane
  • Review + Determination modelsDemonstrated
Artifacts are portableAll layers
  • Export fixtureDemonstrated
  • Rust verifierImplemented
Programs can evolveRails Evidence Plane
  • ProgramProfile versionsDemonstrated
Buyers will pay for relianceAll layers
  • Pricing researchResearched
Adversarial matrix →Wave 0 technical validation →

Why integrations must get cheaper

The infrastructure thesis fails if every customer becomes a software project.

The first deployment can contain learning. The second should reuse primitives, fixtures, adapter patterns, program structures and deployment procedures. The third should reuse more. Measure new core code, unique connector work, profile reuse, founder hours, deployment hours and time to first useful artifact.

Declining marginal implementation work is the technical form of venture scalability.