Skip to content

README.md

Latest commit

 

History

History
107 lines (76 loc) · 4.35 KB

README.md

File metadata and controls

107 lines (76 loc) · 4.35 KB
Raw

GridKeep

GridKeep is a private, original dependency intelligence evaluation stack for the infrastructure beneath communications and cloud. The v0 demo is built around one operator question:

If this node goes down, what breaks, what reroutes, what does not, and what delays restoration?

Tagline: Protect the network beneath the network.

What v0 Proves

  • One Oregon continuity-corridor story, not a generic globe.
  • A strict Output -> Claim -> Evidence trust path.
  • A reduced ontology centered on landing-side communications continuity.
  • A deterministic scenario runner with explainable reroute and restoration logic.
  • A Dockerized single-app eval stack with no auth, no external DB, and bounded live public-context ingestion.
  • Explicit tradecraft and governance surfaces for source quality, uncertainty, alternatives, and CAI/PAI handling.

Repo Shape

  • schema/v0.yaml Human-authored source-of-truth ontology and contract spec.
  • packages/contracts Schema compiler, generated JSON Schemas, generated TypeScript types, seed bundle, and validators.
  • apps/eval Next.js operator console, API routes, deterministic scenario engine, and CLI runner.
  • docs/founding-memo.md Founding memo for the concept and market wedge.

Default Story

The bundled story centers on an Oregon landing-side dependency near Pacific City feeding a Hillsboro connectivity corridor and AWS US West (Oregon).

The default scenario proves all of the following:

  • a modest-looking node fails
  • one downstream service degrades first
  • one dependency reroutes successfully
  • one apparently viable alternate is rejected for shared risk or insufficient headroom
  • restoration is delayed by non-network constraints including permit, utility, supplier, depot, or vessel availability
  • every visible conclusion resolves to claims and evidence

Run Locally

npm install
npm run schema:build
npm test
npm run build
npm run scenario:run
npm run dev

The app runs at http://localhost:3000.

Run With Docker

docker compose up --build

If Compose is unavailable on the host, npm run docker:up falls back to docker build plus docker run automatically.

That brings up the full private evaluation stack at http://localhost:3000.

API Surface

  • GET /api/bootstrap
  • POST /api/run-scenario
  • GET /api/claim/[id]
  • GET /api/evidence/[id]

Core Principles

  • Keep this concept isolated from other workspace projects.
  • Treat provenance as product infrastructure, not a citation afterthought.
  • Prefer corridor-level certainty over invented facility-level precision.
  • Keep claims meaningful rather than ceremonial.
  • Keep explanation more important than score.
  • Keep machine outputs separate from the hard claim graph.
  • Keep privacy and civil-liberties constraints visible whenever public or commercially available context is used.

Standards Alignment

See docs/odni-standards-alignment.md for the concrete GridKeep mapping.

Open Core vs. Full Platform

This repository contains the open core: the ontology, schema, provenance engine, deterministic scenario runner, and baseline console. Everything needed to inspect the model, verify the evidence chain, run a single-corridor proof, and validate claims end to end.

The full platform — including expanded corridor and chokepoint analysis, live multi-source telemetry ingestion, the immersive global theater, premium data adapters, operational deployment tooling, and analyst workflow — is maintained in a separate distribution available under commercial license.

For access to the full platform, contact the GridKeep team.

Non-Goals

  • People surveillance
  • Person-level targeting or pattern-of-life monitoring
  • Generic cyber threat feed aggregation
  • Live ingest pipelines
  • Collaboration workflows
  • Auth and permissions work
  • Global map exploration as the primary experience
  • Cross-project branding, data, or narrative reuse