__ __ ___ ___ _ \ \ / / | __|| _ \ /_\ \ V / | _| | / / _ \ \_/ |___||_|_\/_/ \_\
# Vertex Execution Routing Adapter — codex cli + claude code, one endpoint
Run your existing CLI harness against Vertex AI Gemini without rewriting a single prompt.
Vera terminates the OpenAI Responses and Anthropic Messages wire protocols at vera.5pm.ai, signs requests with its Vertex service account, and streams Gemini back. Every call is metered: tokens in, out, thinking, cached — latency, dollars.
$ awaiting input
harnesses
2 / 2
codex-cli ≥0.157 · claude-code ≥2.1
gemini models served
5
3.8-flash default · /model to switch
metered fields / request
16
tokens, thinking, cached, ttft, usd, …
install
┌─ sh ────────────────────────────────────────────────────────────────────┐ │ curl -fsSL https://vera.5pm.ai/install.sh | sh -s -- --key vera_sk_... │ └──────────────────────────────────────────────────────────────────────────┘
# then open a new shell: vera-codex / vera-claude route through vertex; plain codex / claude stay untouched. uninstall: append uninstall.
$ awaiting input