Agent Management

Framework-Agnostic Agents

Define agents once in Eevee Studio or wrap existing ones. Deploy across LangChain, CrewAI, DSPy, or any framework with consistent governance and lifecycle management.

Unified Lifecycle Events

Eevee normalizes agent events like prompt input, tool calls, and responses across all frameworks. Apply the same governance regardless of the underlying stack.

Framework-agnostic event hooks
Real-time monitoring and observability
Automatic policy enforcement at each stage

Centralized Definitions

Store agent configurations, prompts, tools, and knowledge in one place. Version control and deploy changes without code updates.

YAML/JSON agent manifests
Version history and rollback
Hot reload in production

Wrap or Load

Two deployment patterns: wrap existing agents with Eevee SDK for instant governance, or load agent definitions from Eevee Server for full lifecycle control.

Zero-config wrapping with ev.with_ev()
Dynamic loading from Eevee Server
Works with any Python agent framework

Team Collaboration

Non-developers can create and modify agents in Eevee Studio while developers integrate them seamlessly in code. Everyone works on the same definitions.

Visual agent builder in Studio
Real-time testing and iteration
No deployment required for updates

Two ways to deploy agents

Wrap existing agents or load from Eevee Server - both get full governance

wrap_agent.py

import eevee as ev
from langchain import Agent

# Wrap existing agent
my_agent = Agent(
    tools=[...],
    prompt="..."
)

# Add governance instantly
secure_agent = ev.with_ev(
    my_agent,
    policies=["block_pii"]
)

secure_agent.run()

load_agent.py

import eevee as ev

# Load from Eevee Server
agent = ev.load(
    "agent",
    "customer_support"
)

# Complete definition loaded:
# - Prompts
# - Tools  
# - Policies
# - Knowledge

agent.run()

Start building governed agents

Deploy agents across any framework with centralized control and compliance