◆ novafabric v0.31 · experimental

The time machine
for AI systems.

Capture, replay, and audit AI-agent executions. Evidence infrastructure for reproducible AI. Local-first, cryptographically signed.

nova — bash
$ nova capture python agent.py
capsule ─ 4f8a1c2e
trace.jsonl 1,203 spans
model-calls.jsonl 6 LLM calls
dsse signature ed25519
capsule sealed ◆

01

Your agent ran for 3 hours.

It failed at step 47. You have no idea why.

02

nova capture wraps any command.

Every tool call, model interaction, and environment state — recorded.

$ nova capture python agent.py

03

The capsule holds everything.

Signed. Timestamped. Replayable. Forever.

  • trace.jsonl ✓ 1,203 spans
  • model-calls.jsonl ✓ 6 LLM calls
  • dsse signature ✓ ed25519

04

Replay. Validate. Audit.

Reproduce any run. Verify any claim. Evidence, not trust.

$ nova replay --capsule 4f8a1c2e --mode forensic
research/
active areas

What the lab
is building.

Six active research and engineering areas, each labeled with its current maturity. Labels come from the NovaFabric source-of-truth docs and are updated only when evidence changes.

no capability uses stable until v1.0 schema freeze

experimental

Evidence + replay layer

Local-first run capsules for capture, replay, diff, validation, and audit workflows. CLI capture, SDK decorator, API/MCP proxy, and OTel GenAI semconv all captured.

capturereplaydiffaudit
experimental

NovaSeal trust layer

Tamper-evident DSSE/RFC 3161 sealing in local profile. Cryptographic SoD enforced. Enterprise HSM/X.509 PKI not claimed.

signingDSSERFC 3161
experimentalprototype

Lineage and provenance

SQLite lineage is experimental (10M-edge KuzuDB benchmark done). KuzuDB backend is prototype. Lineage federation is experimental with legal review pending.

lineageKuzuDBprovenance
all research areas →
capsules/
in the wild

Real runs.
Real capsules.

Each entry is a real project using novafabric to capture and replay AI workflows. The capsule snippet shows what a typical run produces.

submit your capsule →
01

novafabric eval suite

Capturing the novafabric test pipeline itself — every pytest run produces a signed capsule with model calls, tool exchanges, and a redaction proof.

pytestintegrationself-hosted
view repo ↗
$ nova capture python -m pytest tests/integration/
capsule ─ 4f8a1c2e
trace.jsonl ✓ 1,203 spans
model-calls.jsonl ✓ 6 LLM calls
dsse signature ✓
02

agent experiment · lineage at scale

A research agent that queries KuzuDB lineage graphs and generates provenance reports. Capsules let us replay any failed query with mocked LLM responses.

KuzuDBlineageagent
view repo ↗
$ nova capture python lineage_agent.py --depth 4
capsule ─ 7b3d9e11
trace.jsonl ✓ 4,847 spans
model-calls.jsonl ✓ 23 LLM calls
mcp-exchanges/ ✓ 14 tool calls
03

HPC job orchestration

SLURM prolog/epilog hooks wrapped with nova capture. Every job submission produces a capsule; failed jobs are replayed in forensic mode without re-running on the cluster.

SLURMHPCforensic
view repo ↗
$ nova capture sbatch --wrap 'python train.py'
capsule ─ c2f0a88d
environment.json ✓ SLURM env vars
trace.jsonl ✓ 892 spans
replay.yaml ✓ forensic mode
Using novafabric on your project?

Open a PR on GitHub — add your capsule snippet to the showcase.

packages/
standalone tools

Standalone tools
in the ecosystem.

nova-worm-conformance
Python · 0.1.0
stable

Behavioral test suite verifying object storage backends meet WORM requirements. 10 mandatory test cases covering immutability, retention locks, legal holds, and lifecycle exemptions.

SEC 17a-4MiFID IICFTC 1.31FINRA 4370
$pip install nova-worm-conformance
nova-collector
Go · 0.1.0-alpha
alpha

Cluster-scale OpenTelemetry collector for capsule events, spans, and metrics. Deployable to Kubernetes via DaemonSet or SLURM via prolog/epilog. Handles millions of events per node.

OpenTelemetryKubernetesSLURMHPC
$go install github.com/novafabric/nova-collector@latest
nova-dashboard
TypeScript · experimental
experimental

Web UI for capsule inspection, lineage visualization, and asset registry browsing. Ships embedded in nova serve. 13 tabs covering all major primitives.

React 19Astro 5Tailwind v4
$nova serve --experimental
about/
mission & principles

One lab.
The full stack of agentic AI.

The next decade of computing will be defined by AI agents operating at the scale of HPC clusters — running Claude and frontier models on thousands of nodes, coordinating across scientific workflows, and making consequential decisions autonomously.

We're building the infrastructure layer for that world: tools for capturing executions, monitoring fleets, governing behavior, and making every agent run reproducible, auditable, and trustworthy.

Open research. Apache-2.0. One engineer. No VC. No data collection.

Portable by design

Capsules are self-contained directories. No proprietary formats, no vendor lock-in. A capsule you create today will be readable in a decade.

Non-invasive capture

Patch SDK call sites and HTTP transports at runtime. No code changes, no decorators required. Works with any Python AI framework.

Evidence-grade integrity

NovaSeal uses DSSE + RFC 3161 timestamps. Verification is always offline and deterministic. Meets financial and clinical compliance requirements.

Scale from laptop to cluster

SQLite for local development. Postgres + RBAC for multi-user server mode. Go collector for SLURM/Kubernetes cluster deployments.

roadmap
v0.1·Asset Registry
v0.2·Run Capsule + Capture
v0.3·Replay + Diff
v0.4·Lineage Graph
v0.5·MCP Capture
v0.6·Wire-level + Multi-target
v0.7·Server Mode + RBAC
v0.8·Policy + Approval Gates
v0.9·Standard Eval Suites
v0.10·NovaSeal + KuzuDB v2
v0.11·Dashboard (13 tabs)
v0.12·Asset Intelligence
v0.13·Maker-Checker Dual Approval
v0.14·NovaSeal Chain Signing
v0.15·Distributed Capsule Runtime
v0.16·Governance + Audit Framework
v0.17·Evidence Fabric + 3D Topology
v0.18–0.19·Dashboard Parity
v0.20–0.21·ClickHouse Cost + Topology
v0.22–0.23·Ed25519 + OAS v1.0 Spec
v0.24·B-Tier Feature Completeness
v1.0·Schema freeze + Production