Caveman Proxy
Byte-safe gateway. Truthful spend, down to the byte
A transparent LLM reverse proxy that measures truthful spend and telemetry, adds provider-native cache hints to the upstream request only (never altering model-visible bytes — record mode is always pass-through), and feeds the Cave Architect profiler that turns that telemetry into a ranked, dollar-quantified plan. Self-hostable; the managed form is Caveman Cloud. In private development.
The byte-safe request path
The proxy is a drop-in base URL. It reads every request, adds provider-native cache hints to the upstream request only, and never alters the bytes your model sees. In record mode it is pure pass-through.
What it changes is what you can see: truthful spend and telemetry, priced to the cent — the input the Cave Architect turns into a ranked plan.
telemetry
- model
- claude · gpt · gemini
- tokens in / out
- 1,284 / 312
- cache hint
- upstream only
- priced
- to the cent
- mode
- record · pass-through
in development · no savings are claimed before they are measured on real traffic
What it can do
- Drop-in base URL for any provider — agents don't change
- Truthful per-request spend + telemetry, priced to the cent
- Byte-safe: provider-native hints on the upstream request only
- Cave Architect profiler → ranked, dollar-quantified Cave Plan
- Eval-gated rollout with automatic rollback before anything ships
- language
- Go
- license
- —