API reference — Cyber AI

Base URL

Point any OpenAI-compatible client at https://api.cyberai.prashantdangi.com/v1. Chat lives at /chat/completions under that host.

Auth

Send Authorization: Bearer $CYBERAI_API_KEY. Create the key in the console. The environment variable name used in the guides is CYBERAI_API_KEY.

Chat completions

The body matches the chat completions shape: a lineup id and a messages array. A system message is optional.

request.sh

curl https://api.cyberai.prashantdangi.com/v1/chat/completions \
  -H "Authorization: Bearer $CYBERAI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "cai-glm",
    "messages": [
      {"role": "user", "content": "List the first checks for a burst of failed step-up events."}
    ]
  }'

Lineup ids

cai-glm — GLM 5.3, Z.ai
cai-deephat — DeepHat V1, Kindo
cai-foundation — Foundation-Sec, Cisco
cai-lily — Lily, Security assistant
cai-whiterabbit — WhiteRabbitNeo, DeepSeek-Coder

Errors

These are the statuses the host will use once it is connected.

401 — missing or unknown key.
402 — balance is empty, so the request does not run.
400 — the model id is not in the lineup, or the body is not chat completions JSON.

Retention

The intended policy is that prompt text is not kept after the response is sent. Today the browser never uploads the prompt, because the host is not connected.