Control your AI spend without changing how your team works.
Opviera sits between your people and the AI providers. Your engineers keep Claude Code, Codex, and OpenCode exactly as they are, and one command points them at your gateway. If you would rather use ours, Opviera CLI and the Opviera Harness desktop app come connected already. Either way, every request from that moment on is budgeted, attributed, and recorded.
Control in three simple steps
No new tools for your engineers to learn. It’s a one-line endpoint change.
Connect a client
Point Claude Code, Codex, or OpenCode at your Opviera gateway, or reach for Opviera CLI or Opviera Harness, which come connected already. Hand each person a vsk_… key. That’s the whole setup.
Set policies in the dashboard
Define budgets, allowed/forced models, and per-project rules with simple toggles. No code, no config files.
Watch usage in real time
Track spend per user and project, get monthly invoices, and surface abuse signals before they cost you.
Already using one of these? One command points it at Opviera. It prompts for the developer’s
vsk_… key and writes the config for you.
curl -fsSL https://console.opviera.ai/config/claude | bashcurl -fsSL https://console.opviera.ai/config/codex | bashcurl -fsSL https://console.opviera.ai/config/opencode | bashOpviera CLI and Opviera Harness
Nothing to point anywhere. Both come connected to your organisation’s gateway, so there is no separate model subscription to manage and no way to route work somewhere unbilled.
Opviera CLI
An agentic coding tool for your terminal. One command installs it. Then run opviera in your project and enter your vsk_… key and project once. The key is checked against the gateway before the agent starts, so a bad key fails right away instead of halfway through a task.
curl -fsSL https://opviera.ai/install | bashmacOS, Linux and Windows. In CI, pass OPVIERA_API_KEY through the environment instead of typing it. It gets validated the same way. Read the docs ›
Opviera Harness
The desktop app, for the people on your team who don’t live in a terminal. Point it at a folder of documents, spreadsheets, notes, a website or code, and it works on the files inside it. Before anything substantial it tells you what it means to do and waits for you to say yes. That folder is the boundary, and it can’t see or change anything outside it.
There is nothing to configure. Install it, paste the key your administrator gave you, pick a folder, and ask.
macOS and Windows. The Mac builds are signed and notarized by Apple. Download it ›
It makes no difference which one a request comes from. The same budgets, the same model rules and the same records apply to all five, and the dashboard tags every request with the client that made it.
What happens on every request
Your rules are applied before a request ever reaches a model, so a budget is a hard stop rather than a report you read afterwards.
-
1 · Identify
Who is asking, and for which project
The developer’s
vsk_…key identifies the person, and the project tag identifies the work. Every request is attributed before it runs, not reconstructed from an invoice later. -
2 · Check the budget
Project cap, then personal quota
The project’s spend cap is checked first, then the developer’s daily and weekly limits on tokens, dollars, and sessions. Over the line, the request is refused with a clear message instead of being billed.
-
3 · Apply your model rules
The model you allow, not the one the CLI picks
If the CLI asks for a model you have not allowed, Opviera serves the best model you have allowed instead of failing. Turn on smart routing and routine work is served by a lighter, cheaper model automatically.
-
4 · Forward & record
Real cost, on the dashboard, right away
The request goes to the provider, the response streams straight back to the developer, and the exact token counts and cost land on the dashboard. Failed requests are never billed, and cache savings are passed through.