One key for every coding agent
Set up your tool
| Tool | Where to point it |
|---|---|
| Claude Code | ANTHROPIC_BASE_URL=https://api.digitspace.ai and ANTHROPIC_AUTH_TOKEN=your key |
| Cursor | Settings → Models: override the OpenAI base URL with https://api.digitspace.ai/v1 |
| Cline | API provider "OpenAI Compatible", base URL https://api.digitspace.ai/v1 |
| OpenCode | An OpenAI-compatible provider with baseURL https://api.digitspace.ai/v1 |
| Aider | --openai-api-base https://api.digitspace.ai/v1 --model openai/auto |

In 1953 an operator put each call through by hand. DigitSpace routes every request.
Switchboard operators, NIH Clinical Center, 1953. NIH History Office, public domain, via Wikimedia Commons.
Why agents in particular
- Long runs survive outages
- An agent loop doesn't stop because one provider returns a 529; the request moves on.
- Cost per tool
- Where a tool names itself in its requests, its spend shows up separately, so you can see what each tool cost.
- A hard stop
- A key budget per developer or per day ends a runaway loop before the bill does.
- The right model per step
- "auto:best" for planning, "auto:economy" for the small steps a tool sends in between.
Log in from the command line
Tools that support it can open DigitSpace in your browser and receive a key after you approve, so nothing needs copying. Those keys expire after 90 days, so a key left on an old laptop stops working on its own.
Questions
Does Claude Code work with non-Claude models?
Yes. Set ANTHROPIC_MODEL to any model, or to auto; DigitSpace translates between the Anthropic and OpenAI formats.
How much does a month of agent use cost?
It depends on the model and how hard you use it; the coding agent cost calculator estimates it from your own numbers.
Can each developer have their own key?
Yes, with their own budget, so team spend is visible per person.
Read next
Live in a minute.
Create a key, change the base URL, send a request. No card needed with your own provider keys.
Start free