API reference

Estro speaks the OpenAI chat-completions dialect. Point any client, SDK, or IDE extension at it by changing the base URL and key.

Estro is available on your favorite platforms:

Pick a platform above to see its setup.

Authentication

Sign in to get a session token. Pass it as Authorization: Bearer <token> or X-Api-Key. Tokens last 30 days and can be revoked from your account page.

Run

curl https://estro.help/api/auth/login \
  -H "Content-Type: application/json" \
  -d '{"username":"you","password":"yourpassword"}'

# Response:
# {"token":"eyJ...","user":{"id":1,"username":"you",...}}

Models

GET /v1/models returns this list live.

Model ID	Name	Best for
`kali-gamma`	Kali Gamma	Deep reasoning, multi-step work, large codebases
`kali-beta`	Kali Beta	Everyday coding, writing, extended context
`kali-epsilon`	Kali Epsilon	Fast iteration, quick answers

Chat completions

POST /v1/chat/completions takes the standard OpenAI request shape. Set stream: true for server-sent events.

Run

curl -N https://estro.help/v1/chat/completions \
  -H "Authorization: Bearer $ESTRO_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "kali-gamma",
    "stream": true,
    "messages": [
      {"role": "user", "content": "Refactor this function and explain the tradeoffs."}
    ]
  }'

Reasoning / chain-of-thought

When a model thinks before answering, stream chunks include reasoning_content deltas before the answer content. The chat UI renders these as a collapsible "Thought process" block.

Run

# Reasoning deltas appear as "reasoning_content" in each chunk's delta.
# The full response includes a collapsible thought process.

curl -N https://estro.help/v1/chat/completions \
  -H "Authorization: Bearer $ESTRO_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "kali-beta",
    "stream": true,
    "messages": [{"role": "user", "content": "Prove that sqrt(2) is irrational."}]
  }'

# Stream events include:
# {"choices":[{"delta":{"reasoning_content":"Let me think..."}}]}
# {"choices":[{"delta":{"content":"Here is the proof..."}}]}

Request format

Field	Type	Required	Description
`model`	string	yes	`kali-gamma`, `kali-beta`, or `kali-epsilon`
`messages`	array	yes	Array of `{role, content}` objects. Roles: `system`, `user`, `assistant`
`stream`	boolean	no	Default `false`. When `true`, response is SSE with `data:` lines
`temperature`	number	no	Sampling temperature, 0 to 2. Passed through to the model
`max_tokens`	integer	no	Maximum tokens in the response
`top_p`	number	no	Nucleus sampling threshold
`stop`	string \| array	no	Stop sequences

Response format

Non-streaming responses return standard chat.completion JSON:

{
  "id": "chatcmpl-abc123",
  "object": "chat.completion",
  "model": "kali-gamma",
  "choices": [{
    "index": 0,
    "message": {
      "role": "assistant",
      "content": "Here is the refactored function...",
      "reasoning_content": "Let me analyze the code..."
    },
    "finish_reason": "stop"
  }],
  "usage": {
    "prompt_tokens": 42,
    "completion_tokens": 128,
    "total_tokens": 170
  }
}

Errors & limits

Status	Code	Meaning
`400`	`bad_request`	Malformed JSON or empty messages
`401`	none	Missing or expired session token
`403`	`email_unverified`	Verify your email address first
`403`	`banned`	Account suspended by admin
`402`	`billing_error`	No active plan. The API and Kali Gamma and Beta need one; see pricing
`404`	`model_not_found`	Unknown model id
`429`	`rate_limit_exceeded`	Daily request cap reached (resets at midnight UTC)
`429`	`usage_limit_reached`	A 5-hour or weekly window is full; `resets_in` and `Retry-After` say when
`502`	none	The model provider is unavailable

Error bodies follow the OpenAI shape: {"error":{"message":"...","type":"invalid_request_error","code":"..."}}

All endpoints

Chat & models

Method	Path	Auth	Description
`POST`	`/v1/chat/completions`	required	Chat completion (stream or JSON)
`GET`	`/v1/models`	no	List available models
`GET`	`/api/models`	no	Models with availability status
`GET`	`/api/usage`	required	Your daily usage + limit

Keys

Create a long-lived key from your account page or with these calls. A key works anywhere a session token does.

Method	Path	Auth	Description
`POST`	`/api/auth/login`	no	Sign in, returns a session token
`GET`	`/api/account/api-keys`	required	List your keys (prefix only)
`POST`	`/api/account/api-keys`	required	Create a key; the secret is returned once
`DELETE`	`/api/account/api-keys/:id`	required	Revoke a key

Rate limits

Model endpoints allow 120 requests a minute per account with a burst of 120, which is plenty for agent loops that fire many calls in sequence. Sign-in and password endpoints allow 10 a minute per IP. Over the limit you get 429 with a Retry-After header; the daily request cap is separate and resets at midnight UTC.