Anthropic

Anthropic provides chat models only. For embeddings, choose another provider such as OpenAI or Ollama. This page covers the cloud setup, the Anthropic-shaped local server and how the test probe authenticates.

When to use

You want Claude for entity extraction or answers.
You can run embeddings on another provider.

Cloud

export EDGEQUAKE_LLM_PROVIDER=anthropic
export ANTHROPIC_API_KEY=sk-ant-...
export EDGEQUAKE_LLM_MODEL=claude-sonnet-4-6
export EDGEQUAKE_EMBEDDING_PROVIDER=openai   # Anthropic has no embeddings
export OPENAI_API_KEY=sk-...

Variable	Purpose	Default
`ANTHROPIC_API_KEY`	API key. Sent as the `x-api-key` header with `anthropic-version`.	none
`ANTHROPIC_AUTH_TOKEN`	Used when `ANTHROPIC_API_KEY` is empty.	none
`ANTHROPIC_BASE_URL`	Server root for an Anthropic-shaped server	Anthropic cloud
`EDGEQUAKE_LLM_MODEL`	Chat model. The catalog lists `claude-sonnet-4-6` and `claude-opus-4-8`.	client default `claude-sonnet-5-5`

Set EDGEQUAKE_LLM_MODEL explicitly so the model does not depend on the client default.

Anthropic-shaped local server

Any server that implements POST /v1/messages works:

export EDGEQUAKE_LLM_PROVIDER=anthropic
export ANTHROPIC_BASE_URL=http://127.0.0.1:4000   # your server's port
export ANTHROPIC_API_KEY=local

Use the server root with no trailing path. In Docker, use http://host.docker.internal:PORT.

Verify the test probe

POST /api/v1/providers/test with "shape":"anthropic_messages" does this:

Lists models at /v1/models.
Sends a short message to /v1/messages with max_tokens set to 8.

It sends the key as x-api-key by default. Set "auth_scheme":"bearer" to send Authorization: Bearer instead. The probe skips the embedding check for this shape. See Test a server.

%%{init: {"theme":"base","themeVariables":{"primaryColor":"#E0E7FF","primaryBorderColor":"#6366F1","primaryTextColor":"#1E1B4B","secondaryColor":"#D1FAE5","secondaryBorderColor":"#10B981","secondaryTextColor":"#064E3B","tertiaryColor":"#FEF3C7","tertiaryBorderColor":"#F59E0B","tertiaryTextColor":"#6B7A90","lineColor":"#7A889C","clusterBkg":"rgba(99,102,241,0.07)","clusterBorder":"#A5B4FC","noteBkgColor":"#FEF9C3","noteTextColor":"#422006","textColor":"#6B7A90","titleColor":"#6B7A90","signalColor":"#7A889C","signalTextColor":"#6B7A90","loopTextColor":"#6B7A90","edgeLabelBackground":"#F1F5F9","actorLineColor":"#94A3B8"}}}%%
%% eq-theme:v1
sequenceDiagram
  participant U as Operator
  participant EQ as EdgeQuake
  participant AN as Anthropic-shaped server
  U->>EQ: POST /api/v1/providers/test
  EQ->>AN: GET /v1/models (x-api-key)
  AN-->>EQ: model list
  EQ->>AN: POST /v1/messages (max_tokens 8)
  AN-->>EQ: message response
  EQ-->>U: test result

Troubleshoot

401 or 403 from the server. The key is wrong, or you set auth_scheme to bearer for a server that expects x-api-key. Check both.
Probe shows model_not_found. The model name is not in the list returned by /v1/models.
Embedding errors. Anthropic has no embedding endpoint. Set EDGEQUAKE_EMBEDDING_PROVIDER to a provider that has one, such as OpenAI or Ollama.

Notes

A saved Connection for an Anthropic-shaped server uses api_shape=anthropic_messages (see Roles).
For a local setup with no key, see Ollama.