Remember what matters.
Curated memory stores environment facts and user preferences across sessions. Searchable session history helps an agent retrieve earlier work when it needs more context.
How persistent memory works ↗
Talk to us ↗AI AGENTS, BUILT AROUND YOUR WORLD
AI agents.
Your environment.
From a question to coordinated work. Give your agents the context, tools, and infrastructure to move things forward.
Run Nedura in your cloud tenant or on your servers. Connect approved models and business tools, keep useful context across sessions, and reach your agents from the channels your team already uses.
HOW WORK MOVES THROUGH NEDURA
TerminalMessagingAPI / IDE
Nedura runtimeContext → Model → Tool → Response
Memory & skillsSession context
◇ModelsConfigured providers
⌘ToolsBuilt-in + MCP
▤StateSessions & recall
CONCEPTUAL SYSTEM VIEW Deployed in infrastructure you operate
YOUR INFRASTRUCTURECONNECTED MODELS & TOOLSCONTEXT THAT CARRIES FORWARD
One agent platform.
Many ways to work.
Build on contextMemory, skills, session recall
Connect your worldMessaging, MCP, model providers
Coordinate workDelegation, schedules, event triggers
EXPLORE NEDURA
01 / THE PLATFORM
Useful work starts
with the right context.
Nedura combines an agent runtime with persistent memory, reusable skills, and an extensible tool system. Configure it for your environment, then build workflows that can carry forward.
MEMORY · SKILLS · COORDINATION Conceptual illustration
Turn experience into skills.
Skills are knowledge documents loaded when needed. Nedura can create and reuse them, and supports the open Agent Skills standard and external skill directories.
Explore the skills system ↗Give each agent its own home.
Profiles keep configuration, credentials, memory, sessions, skills, and gateway state separate. Use independent profiles for different roles and environments.
Understand profiles ↗02 / INSIDE THE SYSTEM
Follow a request.
See the moving parts.
Entry points connect people and applications to the agent runtime. The runtime assembles context, resolves a provider, calls tools, and returns a response. Explore the layers below.
NEDURA / AGENT ARCHITECTUREOne runtime.
One runtime.
A connected ecosystem.
Context, models, tools, and sessions arranged around a central agent runtime.
PEOPLE & APPLICATIONSStart a requestCLI · Gateway · API / ACP
Memory & skillsRelevant context.
Reusable procedures.
Reusable procedures.
Dispatch tools · Return response Agent orchestration
Model providersYour configured model
and API connection.
and API connection.
Session storageSQLite + searchable history
Tool backendsTerminal · Browser · Web · MCP
01
Illustrative request flowA request arrives
The selected entry point passes a task into the agent runtime.
One runtime. Multiple entry points.
Start from the CLI, a configured messaging platform, an API client, or an ACP-compatible editor. The gateway handles platform events and routes messages into sessions.
Simplified from the Nedura architecture documentation. Connections are conceptual; model and external tool requests follow your configured providers.
03 / CONNECTED CAPABILITIES
Bring the right tools
to every task.
Enable toolsets by platform and extend the runtime through MCP and plugins. Use the connections your workflow needs.
Terminal & files
Run commands, manage processes, read and edit files, and work with configurable terminal backends. Keep execution aligned with your chosen environment.
Tools & toolsets ↗Web & browser
Search the web, extract content, and automate browser tasks. Browser and tool backends are configured independently of the conversation surface.
Browser automation ↗MCP connections
Connect local stdio or remote HTTP tool servers. Discover tools at startup and filter which tools each server exposes to the agent.
MCP integration ↗Model choice
Use configured model providers or compatible endpoints. Configure fallback providers when the primary model encounters errors.
Fallback providers ↗Messaging & voice
Connect Teams, Slack, Telegram, Discord, and other supported platforms through a gateway. Media and voice capabilities vary by platform and configuration.
Messaging capabilities ↗Plugins & extensions
Add tools and lifecycle hooks, or connect alternative memory providers and context engines. Extend the platform without modifying the core runtime.
Explore plugins ↗04 / WORK THAT MOVES FORWARD
Coordinate the work.
Carry back the result.
Break a larger task into focused workstreams, or run a repeatable workflow on a schedule. These illustrations show supported patterns, not live execution.
PARALLEL DELEGATIONANIMATED CONCEPT
Parent agent Goal + relevant context
01ResearchIsolated context
02BuildOwn tools
03ReviewFocused task
Final summaries return to the parent
Delegate without filling the parent’s context.
Child agents work with fresh conversations, inherited tool access, and their own terminal sessions. Their final summaries return to the parent for coordination.
Read about delegation ↗SCHEDULED & EVENT-TRIGGERED WORK
- 01Choose a trigger
A one-time schedule, recurring cron expression, or configured webhook event.
- 02Attach the workflow
Run a fresh agent session with relevant skills, or use script-only mode for a deterministic job.
- 03Deliver the result
Send output to the origin chat, a configured platform target, or local files.
Give repeatable work a rhythm.
Create, pause, resume, edit, and trigger jobs through the scheduling tools. Pick the delivery destination that fits the task.
Explore scheduled tasks ↗05 / CONTROL & GOVERNANCE
Choose the foundation.
Set the boundaries.
Nedura and Nedura Gov are two editions of the same product. Nedura runs in infrastructure you operate. Configure tools, platform access, and approval behavior for your environment. Nedura Gov adds Sevengate governance.
NEDURA
Connected agents
for your organization.
For commercial organizations and business IT teams. Run the agent runtime, memory, skills, and extensible tools in your own environment, and connect to approved model providers, chat tools, and business tools.
- Customer-operated deployment
- Independent profiles and sessions
- Configurable command approvals and toolsets
- Messaging, MCP, and model integrations
NEDURA GOV
Policy-controlled actions.
Traceable decisions.
For defense, federal, and regulated organizations requiring the Sevengate governance layer.
- Agent actions denied unless policy allows them
- Human approval for sensitive actions in a terminal, chat, or Microsoft Teams using Microsoft Entra identity
- Tamper-evident, hash-chained decision records
- Governance built into Nedura Gov
06 / DEPLOYMENT
Your environment.
A clear starting point.
Nedura is installed and upgraded by your administrator using nedura-deploy. The documentation covers Linux, Windows Server, containers / Kubernetes using Helm, and Terraform on Azure deployment paths. The platform is deployed using signed releases.
- 01 / Prepare the environment
Choose the infrastructure your organization will operate and review the installation guidance with your administrator.
- 02 / Configure the connections
Set up model providers, tool access, profiles, and the messaging platforms your team needs.
- 03 / Start with a focused workflow
Use the quickstart, validate your configuration, and add skills or automation as your workflow develops.
GO DEEPER
From first session
to your next workflow.
Nedura is built and supported by ALINEDS, a Texas-based IT, AI, cloud, and cybersecurity company serving defense, federal, civilian, intelligence, and commercial customers.