Deployable Artifact Standard

How guideStone becomes a self-leveling benchmark and portable science — the six-layer artifact model from entry point to plain-text documentation.

From Repository to Portable Science

guideStone certifies that a binary produces reproducible results. The Deployable Artifact Standard extends this to portability: the certified binary, its reference data, its integrity manifest, and its documentation travel as a self-contained object — USB drive, tarball, or OCI container.


Six Layers

LayerPurposeContents
1. Entry pointOne-command accessCHECKSUMS, detection script, platform fallback
2. Binary layoutMulti-arch binariesx86_64 static + GPU, aarch64 static
3. Container fallbackNon-Linux accessOCI container image, Windows launcher
4. IntegrityTrust without trustSHA-256 manifest, CRC payloads, Merkle when provenance trio wired
5. Self-knowledgeWhere has this been?liveSpore.json — tracks every machine visited
6. DocumentationHuman-readable contextReference papers, tolerance derivations, plain-text README

Self-Leveling Benchmark

The validation IS the benchmark. When ./hotspring validate runs on unknown hardware:

  1. Detect CPU architecture (x86_64, aarch64)
  2. Probe GPU adapters (Vulkan, fallback to CPU)
  3. Run physics checks (59 checks in the first artifact)
  4. Report pass/fail, wall time, throughput, GPU utilization

The consumer gets two answers from one command: Is the physics correct? and How fast is this machine? No installation, no configuration, no dependency resolution.


Cross-Platform Matrix

PlatformMethodDependencies
Linux x86_64Native binaryNone (static musl)
Linux aarch64Native binaryNone (static musl)
HPC / Slurmsrun ./hotspring validateNone
Windowshotspring.bat (WSL2 -> Docker fallback)WSL2 or Docker Desktop
macOSOCI containerDocker or Podman
CI/CDContainer imageAny OCI runtime

liveSpore.json — Self-Knowledge

Each artifact carries a liveSpore.json that records every machine it has visited:

{
  "artifact": "hotSpring-guideStone-v0.7.0",
  "created": "2026-04-10T00:00:00Z",
  "visits": [
    {
      "hostname_hash": "a3f2...",
      "arch": "x86_64",
      "gpu": "NVIDIA RTX 3090",
      "checks_passed": 59,
      "checks_total": 59,
      "wall_time_seconds": 42.3,
      "timestamp": "2026-04-11T14:30:00Z"
    }
  ]
}

The artifact knows where it has been and what it found. A PI reviewing cross-substrate validation does not need to ask “has anyone tested this on AMD?” — they check liveSpore.json.


Relationship to Other Standards

StandardWhat It CertifiesRelationship
ecoBinBinary structureStructure axis
fieldMouseDeployment surfaceDeployment axis
guideStoneOutput verificationVerification axis
Deployable ArtifactPortable deliveryCombines all three

The Deployable Artifact Standard is the intersection: a structurally compliant binary ( ecoBin), verified by guideStone, packaged for any deployment surface ( fieldMouse).


Evolution Targets

TargetStatus
ARM GPU accelerationPlanned (Jetson, Apple Silicon)
genomeBin manifestPlanned (full composition artifact)
BearDog signingPlanned (cryptographic attestation of artifact integrity)
helixVision artifactPlanned (genomics self-leveling benchmark)
Multi-artifact deploy graphsPlanned (compositions of compositions)

The artifact is the conversation. Not a pitch deck. Not a publication. Not a “contact us for a demo.” A self-contained, self-verifying, self-benchmarking object that runs the science on any machine and answers its own questions. Plug it in, run one command, read the results.