Skip to content

Examples

Explore Nexus through hands-on examples that demonstrate key features and real-world use cases.

🚀 Quick Start Examples

  • File Operations

    Learn how to read, write, copy, move, and delete files with optimistic concurrency control.

    View Example

  • Directory Operations

    Master directory management with hierarchical permissions and automatic inheritance.

    View Example

  • Agentic Context Engineering (ACE)

    Enable AI agents to learn from experience through self-reflection and memory.

    View Example

  • DeepAgents Integration

    Build autonomous agents with event-driven workflows that automatically process outputs and consolidate knowledge.

    View Example

  • LangGraph Integration

    Build ReAct agents with LangGraph using Nexus as persistent filesystem infrastructure.

    View Example

  • LangGraph Migration Tutorial

    Step-by-step guide: Migrate from standard file I/O to Nexus with permissions, cloud storage, and audit trails. See exact code changes needed!

    View Tutorial

  • CrewAI Integration

    Build collaborative multi-agent teams with CrewAI using Nexus for memory, storage, and coordination.

    View Example

  • Claude Agent SDK

    Build intelligent agents with Anthropic's Claude Agent SDK - Built-in ReAct loop with persistent memory.

    View Example

  • OpenAI Agents SDK Integration

    Build production-ready ReAct agents with OpenAI Agents SDK - 70% less code than manual ReAct loops.

    View Example

  • Google ADK Integration

    Build production agents with Google's Agent Development Kit - Multi-agent coordination with session management.

    View Example

  • Permission Management

    Set up fine-grained access control with Google Zanzibar-style ReBAC permissions.

    View Example

  • Workspace & Sessions

    Build multi-zone applications with workspace isolation and session tracking.

    View Example

  • Skill Seekers

    Auto-generate Claude skills from any documentation URL with AI enhancement and llms.txt support.

    View Example

💡 Use Case Examples

Store and query agent context with semantic search.

import nexus
import json

# Connect in embedded mode (no auth needed)
nx = nexus.connect(config={"data_dir": "./nexus-data"})

# Store conversation
conversation = {
    "user": "What are your pricing tiers?",
    "assistant": "We offer Basic ($10/mo), Pro ($50/mo), Enterprise (custom)",
    "timestamp": "2024-01-15T10:30:00Z"
}

nx.write(
    "/agent/memory/conversation.json",
    json.dumps(conversation).encode(),
    metadata={"agent_id": "gpt-4", "session": "abc123"}
)

# Query semantic memory
results = nx.search("/agent/memory", query="pricing preferences")
print(f"Found {len(results)} relevant conversations")

Complete tenant isolation with automatic permissions.

import nexus

# Admin creates tenant workspace
nx = nexus.connect(remote_url="https://nexus.example.com", api_key="admin-key")

nx.workspace.create(
    "/tenant/acme-corp",
    zone_id="acme-123",
    metadata={"company": "Acme Corp", "plan": "enterprise"}
)

# Grant tenant admin permissions
nx.rebac_create("user", "admin@acme.com", "owner", "file", "/tenant/acme-corp")

# User writes to their tenant workspace
user_nx = nexus.connect(remote_url="https://nexus.example.com", api_key="user-key")
user_nx.write("/tenant/acme-corp/data.json", b'{"records": 1000}')

Track every change with built-in versioning.

import nexus

nx = nexus.connect(config={"data_dir": "./nexus-data"})

# Write initial version
nx.write("/model/weights.pkl", b"version 1 data")
nx.write("/model/weights.pkl", b"version 2 data")
nx.write("/model/weights.pkl", b"version 3 data")

# View history
versions = nx.versions.history("/model/weights.pkl")
for v in versions:
    print(f"Version {v.version_number} at {v.timestamp}")

# Roll back
nx.versions.rollback("/model/weights.pkl", version=2)

# Create workspace snapshot
snapshot = nx.workspace.snapshot("/project", name="before-refactor")

Find files by meaning, not just name.

import nexus

nx = nexus.connect(config={"data_dir": "./nexus-data"})

# Store documents (automatically indexed)
docs = [
    "Machine learning improves model accuracy",
    "Deep neural networks for image classification",
    "Natural language processing with transformers"
]

for i, doc in enumerate(docs):
    nx.write(f"/docs/doc{i}.txt", doc.encode())

# Semantic search
results = nx.search("/docs", query="AI and computer vision", limit=5)

for result in results:
    print(f"{result.path}: {result.score}")

📚 Interactive Demos

All examples include runnable shell scripts that demonstrate the full workflow:

Example Script What It Demonstrates
File Operations examples/cli/file_operations_demo.sh Write, read, copy, move, delete with metadata
Directory Operations examples/cli/directory_operations_demo.sh Create directories with permission inheritance
Permissions examples/cli/permissions_demo_enhanced.sh ReBAC permissions, groups, and inheritance
Workspace & Sessions examples/cli/workspace_session_demo.sh Multi-zone isolation and session tracking
Advanced Usage examples/cli/advanced_usage_demo.sh Mounts, versioning, and advanced features
ACE Learning Agent examples/ace/demo_3_data_validator.py Agent learns validation rules from experience
DeepAgents Workflows examples/deepagents/research/demo_2_workflows.py Autonomous agents with event-driven memory consolidation
LangGraph ReAct Agent examples/langgraph/langgraph_react_demo.py ReAct agent with file operations for code analysis
CrewAI Multi-Agent Teams examples/crewai/crewai_nexus_demo.py Collaborative agent teams with persistent memory and coordination
Claude Agent SDK examples/claude_agent_sdk/claude_agent_react_demo.py Intelligent agents with built-in ReAct loop and Nexus tools
OpenAI Agents SDK examples/openai_agents/openai_agent_react_demo.py Production ReAct agents with automatic loop and memory
Google ADK Agent examples/google_adk/basic_adk_agent.py Production agents with session management and multi-agent coordination
Skill Seekers nexus-plugin-skill-seekers/examples/skill_seekers_demo.sh Auto-generate skills from documentation with AI

🎯 What's Next?