API reference

Use a project API key for OpenAI-compatible Chat Completions. Console cookie sessions and API Bearer authentication are separate.

Connection

Setting	Value
Base URL	https://zeroclave.com/v1
Chat Completions	POST https://zeroclave.com/v1/chat/completions
Authorization	Bearer $ZEROCLAVE_API_KEY

Sign in to the console for project-specific endpoints, key management and available models. This documentation reads no session or private project data. Set ZEROCLAVE_API_KEY and ZEROCLAVE_MODEL; use a currently available model ID from the console.

Request parameters

Parameter	Description
model	Required; currently available model ID.
messages	Required; array of conversation messages.
stream	Optional; true enables SSE.
max_tokens / max_completion_tokens	Optional; must be positive integers if supplied, not null or zero. Support and limits depend on the model.
temperature, top_p, tools	Optional; support and allowed values depend on the model.

Examples

curl

curl https://zeroclave.com/v1/chat/completions \
  -H "Authorization: Bearer $ZEROCLAVE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"<MODEL_ID>","messages":[{"role":"user","content":"Hello"}]}'

Python

import os
from openai import OpenAI

client = OpenAI(
    base_url="https://zeroclave.com/v1",
    api_key=os.environ["ZEROCLAVE_API_KEY"],
    max_retries=0,
)
response = client.chat.completions.create(
    model=os.environ["ZEROCLAVE_MODEL"],
    messages=[{"role": "user", "content": "Hello"}],
)
print(response.choices[0].message.content)

Node.js

import OpenAI from 'openai';

const client = new OpenAI({
  baseURL: 'https://zeroclave.com/v1',
  apiKey: process.env.ZEROCLAVE_API_KEY,
  maxRetries: 0,
});
const response = await client.chat.completions.create({
  model: process.env.ZEROCLAVE_MODEL,
  messages: [{ role: 'user', content: 'Hello' }],
});
console.log(response.choices[0].message.content);

Replace <MODEL_ID> in the curl example. Install the openai package for Python or Node.js. SDK automatic retries are disabled here; decide retries based on outcomes and business side effects.

Streaming requests

with client.chat.completions.create(
    model=os.environ["ZEROCLAVE_MODEL"],
    messages=[{"role": "user", "content": "Hello"}],
    stream=True,
) as stream:
    for chunk in stream:
        if chunk.choices:
            print(chunk.choices[0].delta.content or "", end="", flush=True)

This uses the Python client above. Generation can fail mid-stream; read Streaming and Errors.

Models and quota

curl https://zeroclave.com/v1/models -H "Authorization: Bearer $ZEROCLAVE_API_KEY"
curl https://zeroclave.com/v1/quota -H "Authorization: Bearer $ZEROCLAVE_API_KEY"

Models, capabilities and quota depend on the current project and service configuration. Do not rely on historical model lists or pricing in PDFs.

Limits and troubleshooting

Public 402 means insufficient ZeroClave account or project quota; 429 means ZeroClave caller rate limiting. Model service failures use 502/503/504 model_unavailable and do not imply an invalid user key or insufficient user quota. HTTP 503 can also carry pii_mapping_saturated: stop automatic retries and contact an administrator or support with X-Request-ID. This capacity error does not include Retry-After. For ordinary rate limits, respect Retry-After when available; it is a response header, not a JSON body field. See Error Codes.

Privacy boundary

Detection can miss sensitive information. Images, vision inputs and raw files are not covered by the text PII masking guarantee; check what your client actually sends.

Standard API access uses HTTPS. Changing the endpoint and API key does not enable end-to-end encryption or strict remote attestation.

Protection boundaries · Terms of Service