A small, sharp stack
of products.
Each one is a piece of NAVADA Edge taken out, hardened, and shipped. Use them standalone, or together as a single sovereign stack.
NAVADA Edge CLI.
69 commands · 16 tools
The command-line interface for the NAVADA Edge Network. Install once, connect anywhere. Local-first agent that talks to the network over a private mesh or public endpoints, giving you a single scriptable entry point to deployments, automations, agents, memory, and skills.
- · 69 commands covering files, projects, deployments, agents, models
- · 16 built-in tools with a skills framework you can extend in Markdown
- · 3-tier memory — session, project, and persistent
- · 5 AI providers with automatic failover
- · 3-tier response cache — memory LRU, SQLite, vector store
connected · 6 nodes · 29 services
Try:
navada chat — talk to the resident AI
navada deploy — ship a service
navada memory — manage your context
navada skills — extend the agent
navada doctor — health check the mesh
NAVADA Edge SDK.
JavaScript / TypeScript
Programmatic access to the automation queue, edge runtime, and agent workflows. Build internal tools that lean on the network instead of rebuilding the network in your app.
- · Submit, monitor, and cancel automation jobs
- · Stream logs and results back into your app
- · Identity-aware queue access via API key tiers
- · Webhook callbacks on completion and failure
// Submit a job to the automation queue import { NavadaEdge } from 'navada-edge-sdk'; const edge = new NavadaEdge({ apiKey: process.env.NAVADA_KEY }); const job = await edge.queue.submit({ task: 'summarise', input: documents, webhook: 'https://your.app/done' }); console.log(job.id); // "job_a1f3..."
Edge Portal.
Auth · Clerk
Identity-aware control plane for everything you operate on the network — API keys, usage, agent configurations, and audit trail. Cloudflare Zero Trust authentication on every request.
- · Issue and revoke API keys across tiers
- · Real-time usage and quota analytics
- · Configure resident AI agents per workspace
- · Tamper-evident audit log
Operators, developers, and customers
Customers manage their own keys and workloads in a self-service surface, while operators get centralised visibility into every running agent, container, and request — without exposing the underlying mesh.
Edge Compute.
SLA · best-effort
A persistent, always-on task execution surface. Where Edge Compute differs from a Lambda or a Worker is duration: it can run a job for hours or days, hold state, talk to the network, and check in on a schedule.
- · Long-running scrapes, syncs, model fine-tunes
- · Scheduled report generation and delivery
- · Continuous OSINT collection and indexing
- · Webhook fan-out and durable retries
- · Daily competitor scrape across 200+ sites
- · Weekly executive intelligence briefings
- · Continuous social listening and sentiment
- · Quarterly board-pack generation
- · Long-horizon agent simulations
NAVADA Edge Server.
Form factor · Workstation-class
A reference hardware build for organisations that want the entire NAVADA Edge stack in their building, on their network, under their physical control. Workstation-class hardware, pre-imaged, mesh-joined, ready to go.
- · Workstation-grade hardware specification
- · Pre-imaged with the full Edge stack
- · Tailscale mesh + Cloudflare tunnel pre-configured
- · Three-year operational support
Regulated, sovereign, or air-gapped
Healthcare, legal, defence, government — anywhere data residency, jurisdiction, or compliance ruled out the public cloud. Edge Server brings the platform into your perimeter.
NAVADA MCP Server.
Spec · MCP 2025-06
A model context protocol server that exposes eighteen domain tools to any compliant AI agent — Claude, Cursor, OpenAI's tools, your own. Tiered access keys let you give partners, customers, or internal teams differentiated capability.
REQUEST MCP ACCESS →