What problem does it solve?
Keeping two LLM gateway implementations (a Python service and a Go service) in sync is error-prone, and teams need an evidence-based record of which features are supported, blocked, or need verification before migrating callers.
Core Features & Use Cases
- Dual-gateway inspection: Compares routes, auth, billing, rate limiting, providers, models, and event attribution across the Python
services/llm-gateway and the Go PostHog/ai-gateway codebases.
- Evidence classification: Labels each contract as Supported, Blocking, or Verify based on actual implementation code rather than READMEs or existing tables.
- Parity record maintenance: Updates
services/llm-gateway/PARITY.md with source SHAs, verification dates, and migration-relevant contract changes.
- Use Case: When a Go gateway PR changes authentication or billing behavior, run this audit to determine which Python caller classes become eligible or blocked for migration.
Quick Start
Audit the current LLM gateway implementations against each other and refresh the parity record in services/llm-gateway/PARITY.md.