Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.lucid.foundation/llms.txt

Use this file to discover all available pages before exploring further.

​
TrustGate (LLM Gateway)

TrustGate is an OpenAI-compatible language model (LLM) proxy designed to manage tenant authentication, quota, and payment enforcement. It acts as a gateway to ensure secure and efficient access to language models while providing robust tenant management features.

​
Workflow Overview

The request processing in TrustGate follows a structured workflow to ensure that all necessary checks and validations are performed before accessing the language model. Below is a step-by-step breakdown of the process:
  1. Request: The process begins when a request is made to the TrustGate.
  2. Bearer Authentication: The request includes a bearer token, which is authenticated using a SHA-256 key lookup. This step ensures that only authorized users can proceed.
  3. Payment Check (x402): If the tenant has payment enforcement enabled, the system performs a payment check. This ensures that the tenant has fulfilled any necessary payment obligations before accessing the service.
  4. Plan Enforcement: The system checks the tenant’s plan to ensure that the request complies with the terms and conditions of their subscription plan.
  5. Quota Check: The request is evaluated against the tenant’s usage quota. This step ensures that the tenant has not exceeded their allocated usage limits.
  6. Zod Validation: The request undergoes validation using Zod, a schema validation library. This step ensures that the request data is correctly structured and meets the required specifications.
  7. Passport Model Resolution: The system resolves the passport model, which involves determining the appropriate language model and configuration for the request based on the tenant’s profile.
  8. LiteLLM Fetch: The request is forwarded to the LiteLLM, where the language model processes it and generates a response.
  9. Usage Metering: Usage data is recorded in a fire-and-forget manner, ensuring that the tenant’s usage is tracked without impacting the response time.
  10. Receipt Event: A receipt event is generated for Solana anchoring, providing a verifiable record of the transaction.
  11. Response: Finally, the processed response is returned to the requester.

​
Key Features

  • OpenAI Compatibility: TrustGate is designed to be compatible with OpenAI’s language models, making it easy to integrate with existing systems.
  • Tenant Authentication: Secure authentication mechanisms ensure that only authorized users can access the service.
  • Payment and Quota Enforcement: TrustGate enforces payment and usage quotas, providing a fair and sustainable usage model.
  • Schema Validation: Zod validation ensures that requests are well-formed and meet the required specifications.
  • Efficient Usage Tracking: Usage metering is performed efficiently to provide accurate tracking without affecting performance.
  • Solana Anchoring: Receipt events are anchored on Solana, offering a secure and verifiable record of transactions.
TrustGate provides a comprehensive solution for managing access to language models, ensuring security, compliance, and efficient resource utilization.