Platform & SDK
Everything you need to build, deploy, and manage AI agents at scale. From CLI to SDKs to enterprise governance.
Quick start
Install the Nexus CLI and deploy your first agent in under 5 minutes.
$ curl -fsSL https://nexus.run/install.sh | sh Nexus CLI v0.8.0 installed $ nexus init --repo ./my-project Indexing 847 files across 12 modules... Dependency graph built in 1.7s Created nexus.yaml – customize your agent configuration $ nexus deploy --name my-first-agent Deploying agent "my-first-agent"... / Loading model: nexus-1 (default) / Connecting memory layer / Agent online: https://nexus.run/agents/my-first-agent / Status: active • 0 tasks queued
CLI Reference
Every Nexus CLI command, documented with common usage patterns.
| Command | Description | Example |
|---|---|---|
| nexus init | Initialize a project directory and index the codebase | nexus init --repo ./app |
| nexus deploy | Deploy an agent to the Nexus runtime | nexus deploy --name arch --model nexus-1 |
| nexus ask | Ask your agent a question or assign a task | nexus ask "Find performance bottlenecks" |
| nexus apply | Apply a generated plan to the codebase | nexus apply plan --id plan-abc123 |
| nexus status | Check agent status and recent activity | nexus status --agent arch |
| nexus logs | View agent reasoning traces and tool calls | nexus logs --agent arch --tail 50 |
| nexus config | View or modify agent configuration | nexus config set model nexus-1 |
SDK Installation
Choose your language. Every SDK ships with full type definitions, comprehensive documentation, and working examples.
TypeScript SDK
$ npm install @nexus/sdk import { Nexus } from "@nexus/sdk"; const nexus = new Nexus({ apiKey: "..." }); const agent = await nexus.agents.deploy({ name: "architect", repo: "org/backend", model: "nexus-1" }); const result = await agent.ask( "Refactor the auth module to use role-based access" ); console.log(result.plan);
Python SDK
$ pip install nexus-sdk from nexus import Nexus nexus = Nexus(api_key="...") agent = nexus.deploy_agent( name="architect", repo="org/backend" ) result = agent.ask( "Find all deprecated API endpoints" ) print(result.summary)
Agent Configuration (nexus.yaml)
Every agent is defined by a YAML manifest that specifies its goals, tools, memory settings, and guardrails.
# nexus.yaml -- agent configuration example name: code-architect model: nexus-1 context_window: 128k memory: persistent: true retention_days: 90 tools: - compiler - test_runner - code_search - docs_retriever guardrails: require_review: true allowed_paths: ["src/", "tests/"] max_parallel_tasks: 3 deploy: environment: production auto_approve: false notify: ["slack://#engineering", "email://team@org.com"]
Use cases
Teams around the world use Nexus agents across every stage of the software lifecycle. Here are the most common patterns.
/>
Code migration
Deploy an agent to analyze and refactor across entire codebases. Nexus handles dependency analysis, breaking changes, and regression testing.
#
Bug triage & fix
Connect Nexus to your issue tracker. Agents diagnose root causes, generate fixes, run tests, and open PRs with zero human overhead.
~
Code review automation
Agents review every PR for correctness, style, security, and performance. Human reviewers focus on architecture and design decisions.
[ ]
Technical debt management
Schedule agents to continuously scan for deprecated patterns, upgrade dependencies, and generate cleanup plans with risk assessment.
&
Documentation generation
From API reference docs to architecture decision records, Nexus agents generate and maintain living documentation that stays in sync with your code.
::
Onboarding & knowledge transfer
New team members get a personalized agent that knows the codebase, explains architecture decisions, and accelerates ramp time from weeks to days.
Pricing
Simple, usage-based pricing that scales with your team. No hidden fees, no surprise charges.
Starter
Free
For individual developers
- 1 concurrent agent
- Standard tool chain
- Community support
- 7-day memory retention
Pro — Most popular
$49
per developer / month
- 5 concurrent agents
- Full tool chain + custom tools
- Email & chat support
- 90-day memory retention
- CI/CD integration
Enterprise
Custom
For teams that need more
- Unlimited agents
- Self-hosted deployment
- SSO & RBAC
- Unlimited memory retention
- Dedicated support & SLAs
Integrations
Nexus connects directly to your existing toolchain. All integrations are plug-and-play with zero configuration.
FAQ
Platform questions, answered.
What happens when an agent makes a mistake?
Nexus agents are designed with safety in mind. By default, every change requires human review before it is applied. You can configure approval gates, run agents in read-only mode, and set up automated testing to catch errors before they reach production.
Can I run Nexus on my own infrastructure?
Yes. Enterprise plans include self-hosted deployment options. The platform runs on Kubernetes and ships as a single container image with all dependencies included. We provide Terraform modules for AWS, GCP, and Azure.
How does memory retention work across sessions?
Nexus agents maintain persistent memory that survives across sessions, deployments, and restarts. The platform uses a three-tier memory architecture (episodic buffer, compressed working memory, and long-term consolidation) inspired by our published research.
What APIs and SDKs are available?
We ship first-class SDKs for TypeScript, Python, and Go. All SDKs are fully typed and documented. The REST API supports any language. WebSocket connections are available for real-time agent interaction.