Responses

You're Claude Code, working in my repository.

Task
Wrap the Responses endpoints in a small client module for my app: create (POST /v1/responses), retrieve (GET /v1/responses/{response_id}), delete (DELETE /v1/responses/{response_id}), cancel (POST /v1/responses/{response_id}/cancel) and count input tokens (POST /v1/responses/input_tokens), each returning the parsed JSON and raising the API's error message on a 4xx or 5xx.

Read this first
Meta's docs page "Responses" as Markdown: https://dev.meta.ai/docs/api-reference/responses.md
It's the source of truth for request and response shapes. The same page for people: https://dev.meta.ai/docs/api-reference/responses

Setup
- Base URL: https://api.meta.ai/v1
- Auth: send Authorization: Bearer $MODEL_API_KEY, reading MODEL_API_KEY from the environment. Never hard-code, print or commit the key.
- Model: muse-spark-1.3-contributor (Contributor mode: Meta may use what you send to train future models; drop -contributor for Standard mode)
- Python: the OpenAI SDK (pip install openai) with base_url="https://api.meta.ai/v1" and api_key=os.environ["MODEL_API_KEY"] works, or plain requests.

Facts to get right
- Base URL https://api.meta.ai/v1. Responses are stored by default (store: true), so they can be retrieved and chained with previous_response_id.
- Retrieve echoes the request's parameters, usage and completed_at (null until the status is completed).
- Cancel works on background responses only and returns an always-empty output; cancelling one again returns its current state; a deleted one is a 404.
- DELETE soft-deletes and returns {"object": "response.deleted", "deleted": true}; a repeat DELETE is a 404, not a 200.
- input_tokens returns {"object": "response.input_tokens", "input_tokens": n} for the fully rendered input, resolved previous_response_id history included. Nothing is generated or stored.

Starter code (Python)
import os

from openai import OpenAI

client = OpenAI(base_url="https://api.meta.ai/v1", api_key=os.environ["MODEL_API_KEY"])
MODEL = "muse-spark-1.3-contributor"

import openai
import requests

created = client.responses.create(model=MODEL, input="What is the capital of France?")
count = requests.post(
    "https://api.meta.ai/v1/responses/input_tokens",
    headers={"Authorization": f"Bearer {os.environ['MODEL_API_KEY']}"},
    json={"model": MODEL, "previous_response_id": created.id, "input": "And its population?"},
)
count.raise_for_status()
print("follow-up input tokens:", count.json()["input_tokens"])
fetched = client.responses.retrieve(created.id)
print(fetched.status, fetched.completed_at, fetched.usage.output_tokens)
client.responses.delete(created.id)  # returns nothing on success
try:
    client.responses.delete(created.id)
except openai.NotFoundError:
    print("second delete: 404, already gone")

background = client.responses.create(model=MODEL, input="Write a 600-word story.", background=True)
print(client.responses.cancel(background.id).status)

Done when
- One script creates a response, counts a follow-up, retrieves it, deletes it, and shows the 404 on a second delete.
- The key comes from MODEL_API_KEY and never appears in code, logs or commits.
- API errors (4xx, 429) surface their message instead of failing silently.