Home > Products > EulerAgent

EulerAgent

Local-First CLI Agent Framework

A secure agent framework built on an 8-state machine with Human-in-the-Loop (HITL) approval. Pattern and Graph orchestrators, RAG pipeline, Long-Term Memory (SQLite), MCP integration, 30+ CLI commands, and a plugin system. Every action is logged, audited, and approved.

Open Source

Key Features

Safety, Memory, Orchestration -- the three pillars of EulerAgent

Human-in-the-Loop Security

  • Deny-All Security Model: All tools are blocked by default; explicit permission required
  • Human Approval Required: High-risk actions (file writes, shell execution, external calls) require human approval
  • 8-State Machine: CREATE → CONTEXT → PLAN → PROPOSE → APPROVE → EXECUTE → RESULT → FINALIZE
  • Complete Audit Trail: Every action recorded via Audit Trail

Long-Term Memory & RAG

  • SQLite-Based LTM: Persistent conversation history and vector search
  • Automatic Context Extraction: Auto-learns user preferences and project context
  • Hybrid RAG: Local knowledge base + web search with citation tracking
  • MCP Integration: Model Context Protocol for Brave, Tavily, and external tool connectivity

Workflow Orchestration

  • Pattern Orchestrator: YAML-based workflows (12 canonical patterns)
  • Graph Orchestrator: LangGraph-based parallel execution (experimental)
  • Dynamic Task Decomposition: --dynamic option
  • Plugin System: Extensible architecture for custom tools and providers

Agent Templates

Quickly create agents with purpose-built templates

Template Purpose
personal-assistant General-purpose personal assistant
marketing-expert Marketing content creation
code-assistant Code writing and debugging
ops-assistant Operations and infrastructure management

CLI Quick Start

Create and run an agent with a single command

# Initialize workspace
euleragent init

# Create an agent
euleragent new my-assistant --template personal-assistant

# Run in Plan mode (propose then approve)
euleragent run my-assistant --task "Write project summary" --mode plan

# Run in Execute mode (auto-execute low-risk tasks)
euleragent run my-assistant --task "Refactor code" --mode execute

# Approval management
euleragent approve list
euleragent approve accept <id> --actor "user:name" --execute

# Interactive chat (long-term memory enabled)
euleragent chat my-assistant --project default --continue

CLI Reference

30+ commands across 7 groups for complete agent lifecycle management

Core Commands

Command Description
euleragent init Initialize workspace
euleragent new Create agent from template
euleragent run Run agent (plan/execute mode)
euleragent chat Interactive chat session
euleragent logs View execution audit logs
euleragent doctor System health check

Approval Management

Command Description
euleragent approve list List pending approvals
euleragent approve show <id> Approval details
euleragent approve accept <id> Accept and execute approval
euleragent approve reject <id> Reject approval
euleragent approve accept-all Batch approve all

Knowledge Base (RAG)

Command Description
euleragent rag init Initialize knowledge base
euleragent rag add --path ./docs/ Add documents
euleragent rag query --q "search term" Search knowledge base
euleragent rag status Check KB status

Memory Management

Command Description
euleragent memory status Memory store statistics
euleragent memory search --q "search term" Search past conversations

Workflow & Patterns

Command Description
euleragent workflow show <run-id> Workflow structure and status
euleragent workflow resume <run-id> Resume workflow
euleragent pattern list List available patterns
euleragent pattern run <pattern> <agent> Run pattern-based workflow

MCP Management

Command Description
euleragent mcp sync Sync MCP server catalog
euleragent mcp show Current MCP status

Tutorials

Master EulerAgent with step-by-step guides

Tutorials coming soon.

Installation & Getting Started

Install EulerAgent and run your first agent

Installation

pip install euleragent

# Download Ollama model
ollama pull qwen3:32b

# Initialize workspace
euleragent init

Requirements

  • Python 3.11+
  • Ollama (or OpenAI/Gemini API key)

GitHub

eulerwa/euleragent

Configuration Example

Declaratively configure your agent environment with workspace.yaml

workspace.yaml

llm_profiles: local: provider: ollama model: qwen3:32b openai: provider: openai model: gpt-4o-mini is_external: true default_llm_profile: local memory: enabled: true mcp: enabled: true

Start building secure AI agents with EulerAgent

Open source, human-approval-based, local-first execution.

Get Started on GitHub Contact Us