One key for everything the raw LLM API can't do.

# search the marketplace for the capability your API call is missing
curl "https://api.askfaro.com/tools/search?q=generate+an+image+of+a+coffee+mug"

# invoke it with the same Bearer you'd use for anything else on Faro
curl -X POST "https://api.askfaro.com/invoke/<namespace>/<tool>" \
  -H "Authorization: Bearer $FARO_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"arguments": {"prompt": "coffee mug on a wooden table"}}'

import os, requests

FARO = "https://api.askfaro.com"
headers = {"Authorization": f"Bearer {os.environ['FARO_API_KEY']}"}

# your model returns text; Faro covers everything it can't do natively
hits = requests.get(f"{FARO}/tools/search", params={"q": "generate a product image"}).json()
tool = hits["results"][0]["id"]   # e.g. "fal-flux/text-to-image"

out = requests.post(
    f"{FARO}/invoke/{tool}",
    headers=headers,
    json={"arguments": {"prompt": "coffee mug on a wooden table"}},
).json()
print(out["output"])

# one-time install (requires Python 3.11+)
pip install askfaro-cli
faro auth login    # paste your faro_ key

# whenever a prompt needs something the raw API can't do
faro search "narrate this script in a warm female voice"
faro invoke <namespace>/<tool> --params '{"text": "..."}'