• Custom integration development
• Framework-specific optimization
• Enterprise deployment support
• Team training and workshops
Framework Integrations
AgentFacts for Your Favorite Agent Framework
Official and community-supported integrations for popular agentic AI frameworks. Generate verified metadata automatically for your agents with minimal code changes.
8
Official Integrations
15+
Community Projects
100%
Open Source
<5min
Setup Time
Official Integrations
Maintained by the AgentFacts team and framework authors. Production-ready with full support and documentation.
🦜
LangChain
Official
v0.1.0
Generate AgentFacts metadata automatically for LangChain agents. Supports ReAct, conversational, and custom agent patterns.
Python
from langchain.agents import create_react_agent
from agentfacts_langchain import AgentFactsGenerator
# Create your agent as usual
agent = create_react_agent(llm, tools)
# Generate AgentFacts metadata
agentfacts = AgentFactsGenerator.from_agent(
agent=agent,
agent_name="Research Assistant",
compliance={"eu_ai_act": "low_risk"}
)
# Export or register
agentfacts.export("agentfacts.json")
agentfacts.register_blockchain(chain="hedera")
Python
TypeScript
ReAct
Conversational
LangGraph
🚢
CrewAI
Official
v0.1.0
Multi-agent crew verification. Generate AgentFacts for individual agents or entire crews with automatic role-based metadata.
Python
from crewai import Crew, Agent
from agentfacts_crewai import CrewFactsGenerator
# Define your crew
crew = Crew(
agents=[researcher, writer, reviewer],
tasks=[research_task, writing_task, review_task]
)
# Generate AgentFacts for entire crew
crew_facts = CrewFactsGenerator.from_crew(
crew=crew,
crew_name="Content Production Crew"
)
# Or generate for individual agents
agent_facts = CrewFactsGenerator.from_agent(researcher)
Python
Multi-Agent
Role-Based
Hierarchical
🤖
AutoGen
Community
v0.0.9
Microsoft AutoGen integration with support for conversable agents and group chats. Community-maintained with active development.
Python
from autogen import AssistantAgent
from agentfacts_autogen import AutoGenFactsGenerator
# Create your AutoGen agent
assistant = AssistantAgent(
name="research_assistant",
llm_config={"model": "gpt-4"}
)
# Generate AgentFacts
agentfacts = AutoGenFactsGenerator.from_agent(
agent=assistant,
compliance={"nist_ai_rmf": True}
)
agentfacts.export("agentfacts.json")
Python
Conversable Agents
Group Chat
🧠
Semantic Kernel
Official
v0.1.1
Microsoft Semantic Kernel integration supporting C# and Python. Generate metadata for skills, planners, and complete kernel configurations.
C#
using Microsoft.SemanticKernel;
using AgentFacts.SemanticKernel;
// Create kernel
var kernel = Kernel.CreateBuilder()
.AddOpenAIChatCompletion("gpt-4", apiKey)
.Build();
// Generate AgentFacts
var agentFacts = AgentFactsGenerator.FromKernel(
kernel,
agentName: "SK Assistant",
compliance: new { EuAiAct = "low_risk" }
);
await agentFacts.ExportAsync("agentfacts.json");
C#
Python
Skills
Planners
🦙
LlamaIndex
Coming Soon
Q1 2026
RAG-focused integration for LlamaIndex agents. Will support query engines, chat engines, and ReAct agents with automatic data source tracking.
Early Access: Join our waitlist to get notified when LlamaIndex integration launches. We're working with the LlamaIndex team to ensure first-class support.
Python
RAG
Query Engines
ReAct
Community Integrations
Community-maintained integrations for emerging frameworks. Want to add yours? Submit a PR →
Submit Your Integration
Built an AgentFacts integration for your framework? We'd love to feature it here. Submit a pull request with your integration details.
Contribution Guide →Build Your Own Integration
Creating an AgentFacts integration for a new framework is straightforward. Follow this guide to add AgentFacts support to any agent framework.
1
Extract Agent Metadata
Introspect your framework's agent objects to extract relevant information: model, capabilities, configurations, etc.
Python
def extract_metadata(agent):
return {
"model": agent.llm.model_name,
"tools": [tool.name for tool in agent.tools],
"memory": agent.memory.type if agent.memory else None
}2
Map to AgentFacts Schema
Transform extracted metadata into the AgentFacts 10-category structure. Use the JSON schema for validation.
Python
from agentfacts import AgentFactsSchema
def to_agentfacts(metadata, agent_name):
return AgentFactsSchema(
core_identity={
"agent_id": f"did:agentfacts:{agent_name}",
"name": agent_name,
"version": "1.0.0"
},
baseline_model={
"foundation_model": metadata["model"]
},
capabilities={
"tool_calling": True,
"domain_expertise": metadata["tools"]
}
)3
Add Export Methods
Provide convenient methods to export AgentFacts metadata as JSON, register on blockchain, or integrate with governance systems.
Python
class AgentFactsGenerator:
@staticmethod
def from_agent(agent, **kwargs):
metadata = extract_metadata(agent)
return to_agentfacts(metadata, **kwargs)
def export(self, filepath):
with open(filepath, 'w') as f:
json.dump(self.to_dict(), f, indent=2)
def register_blockchain(self, chain="hedera"):
# Blockchain registration logic
passReference Implementation
Study the LangChain integration as a complete reference implementation with tests and documentation.
View Reference Code →Integration Template
Use our cookiecutter template to bootstrap a new framework integration with boilerplate code.
Get Template →Integration Best Practices
✓ Automatic Metadata Generation
Extract as much metadata as possible automatically from the agent object. Minimize required user input.
✓ Compliance Pre-Population
When possible, pre-populate compliance fields based on model characteristics (e.g., EU AI Act risk levels for known models).
✓ Framework-Specific Extensions
Use the extensibility category for framework-specific metadata that doesn't fit the core 10 categories.
✓ Validation & Testing
Validate generated metadata against the AgentFacts JSON schema. Include comprehensive test coverage.
Core API Reference
All framework integrations should implement these core methods for consistency across the ecosystem.
Python API Spec
class AgentFactsGenerator:
"""Base class for framework integrations"""
@staticmethod
def from_agent(agent, agent_name: str, **kwargs) -> AgentFacts:
"""
Generate AgentFacts from framework agent object
Args:
agent: Framework-specific agent instance
agent_name: Human-readable agent identifier
**kwargs: Additional metadata (compliance, performance, etc.)
Returns:
AgentFacts instance with complete metadata
"""
pass
def export(self, filepath: str, format: str = "json") -> None:
"""Export AgentFacts to file"""
pass
def register_blockchain(self, chain: str = "hedera") -> str:
"""Register metadata hash on blockchain"""
pass
def validate(self) -> bool:
"""Validate against AgentFacts JSON schema"""
passFull API Documentation
Complete API reference with examples for all methods and configuration options.
View API Docs →Need Help with Integration?
Jared James Grogan, creator of AgentFacts, provides consulting for custom framework integrations, enterprise deployments, and standards adoption.
Contact: jared.grogan@post.harvard.edu