The Identity Layer for the Open-Source Agentic Ecosystem
The Open-Source Agent Stack Is Missing a Layer
The open-source AI agent ecosystem has never been stronger. LangChain for orchestration. CrewAI for multi-agent coordination. LlamaIndex for data retrieval. AutoGen for agent communication. Llama for open-weight reasoning.
- Who is this agent?
- What is it allowed to do?
- What did it actually do?
- Can we prove it?
Why Frameworks Don't Solve This
Agent frameworks are rightly focused on capabilities — reasoning, tool calling, memory, retrieval. Identity and audit trails are infrastructure concerns that sit below the framework layer.
- Web frameworks (Express, Django) don't ship with user authentication — you add Passport, Auth0, or Clerk
- API frameworks don't ship with rate limiting — you add a gateway
- Database ORMs don't ship with encryption — you add a vault
Agent frameworks don't ship with agent identity. That's what Vorim is.
Making Llama Enterprise-Safe
Meta's Llama models are the most deployed open-weight AI models in the world. Enterprises love the control of self-hosting. But self-hosting means self-governing — and most teams don't have the infrastructure to make autonomous Llama-powered agents accountable.
- Cryptographic identity per agent, regardless of which model it uses
- Permission enforcement at the infrastructure layer, not the model layer
- Signed audit trails that work the same whether you're running Llama, Mistral, or any other model
- Credential delegation so agents can access OAuth services safely
Built-In Integrations
| Framework | TypeScript | Python |
|---|---|---|
| LangChain | @vorim/sdk/integrations/langchain | vorim[langchain] |
| OpenAI | @vorim/sdk/integrations/openai | vorim[openai] |
| Claude | @vorim/sdk/integrations/anthropic | vorim[anthropic] |
| CrewAI | @vorim/sdk/integrations/crewai | vorim[crewai] |
| LlamaIndex | @vorim/sdk/integrations/llamaindex | — |
Each integration wraps tool calls with automatic permission checks and audit trail emission. 3 lines of code to add identity to your existing agents.
The Open Protocol
VAIP (Vorim Agent Identity Protocol) is published as an open specification on GitHub. The IETF Internet-Draft (draft-nyantakyi-vaip-agent-identity) is submitted for standardization.
The protocol is implementation-agnostic. Any framework, any model, any language can implement VAIP. The reference implementations (TypeScript and Python SDKs) are published on npm and PyPI.
Get Started
npm install @vorim/sdk or pip install vorim. Documentation at vorim.ai/docs. Protocol spec at github.com/Vorim-AI-Labs/vorim-protocol. vorim.ai
Ready to build with agent identity?
Free plan: 3 agents, 10K auth events/month, full SDK access. No credit card.