LLM Providers
Add the Anthropic or OpenAI key Bilbis needs for agent work.
What this means
An LLM provider is the AI service Bilbis uses when agents plan, write, review, and explain code changes.
Bilbis can use Anthropic or OpenAI credentials. You can add one or both. The pipeline form can let Bilbis pick an engine automatically, or users can choose a specific engine when they need more control.
When to use it
Add or update an LLM provider when:
- You are setting up a new organization.
- Pipeline creation is blocked by the setup checklist.
- A provider key expired or was revoked.
- You want Bilbis to use a different provider account.
Before you start
You need:
- Owner or admin access in Bilbis.
- An active Anthropic or OpenAI API key.
- Enough provider-side budget or quota for the pipelines your team plans to run.
Supported providers
| Provider | What to know |
|---|---|
| Anthropic | Enables Claude engines such as Opus, Sonnet, and Haiku. |
| OpenAI | Enables OpenAI-backed engines such as Codex. |
Steps
- Open Settings.
- Open Integrations.
- Select Add credential.
- Choose Anthropic or OpenAI.
- Add a label that other admins will recognize.
- Paste the API key.
- Save the credential.
What happens next
Bilbis stores the key encrypted. When a pipeline runs, Bilbis uses the configured provider to power the agents assigned to that pipeline.
If the credential validates successfully, the Integrations page shows it as reachable. If validation fails, the credential is still visible, but pipelines that depend on it may fail until the key is fixed.
Problems and fixes
| Problem | What to check |
|---|---|
| No LLM provider is available | Add an Anthropic or OpenAI credential in Settings -> Integrations. |
| Validation fails | Confirm the key is active and copied exactly. |
| Pipelines stop after a billing change | Check provider-side billing or quota in Anthropic or OpenAI. |
| Users are unsure which engine to pick | Use Auto unless there is a specific cost, speed, or quality reason to choose manually. |