aivisibilityskill

The official Python client for the aivisibilityskill API. Ask ChatGPT, Gemini, Google AI Overviews and AI Mode, Perplexity, Claude and Grok about any brand, in any country, and get back what they said, who they cited and what to fix.

• Safe retries: every check carries an idempotency key, so a retry is never charged twice.
• Webhook signature verification included.
• One dependency (httpx). Python 3.9+. Typed (py.typed).

pip install aivisibilityskill

Quick start

import os
from aivisibilityskill import Aiv

aiv = Aiv(api_key=os.environ["AIV_KEY"])  # aiv_live_... or aiv_test_... (free sample data)

job = aiv.checks.create(
    ["best invoicing app for freelancers"],
    engines=["chatgpt", "gemini", "ai_overview", "perplexity"],
    locale="en-US",
)
done = aiv.jobs.wait(job["id"])
for answer in done["answers"]:
    print(answer["engine"], [b["name"] for b in answer["brands"]])

Get a key at app.aivisibilityskill.com. Test keys (aiv_test_...) are free and return sample answers, so you can build without spending credits.

Building for your own customers

aiv.checks.create(questions, customer="cus_8421", metadata={"workspace": "acme"})

key = aiv.keys.create("acme", customer="cus_8421", allowed_ips=["203.0.113.0/24"])

csv_text = aiv.usage.csv(from_="2026-11-01", to="2026-11-30", group_by="customer")

# Everything stored for a customer, deleted (runs in the background):
deletion = aiv.customers.delete_data("cus_8421")
aiv.data_deletions.wait(deletion["id"])

Thousands of answers at once

batch = aiv.batches.create(
    two_thousand_questions,
    engines=["chatgpt", "ai_overviews"],
    mode="batch",        # cheapest
    max_age_hours=24,    # reuse answers from the last day at half price
)
done = aiv.batches.wait(batch["id"])  # or listen for the batch.completed webhook
for job_id in done["checks"]:
    job = aiv.jobs.get(job_id)

Webhooks

from aivisibilityskill import verify_webhook, WebhookVerificationError

# Flask: request.get_data(), Django: request.body, FastAPI: await request.body()
try:
    event = verify_webhook(raw_body, request.headers, os.environ["AIV_WEBHOOK_SECRET"])
except WebhookVerificationError:
    return "", 400
if event["type"] == "check.completed":
    job = aiv.jobs.get(event["data"]["id"])

During a secret rotation, pass both secrets: verify_webhook(body, headers, [new_secret, old_secret]). Deliveries older than 5 minutes are refused as replays.

Lists

Lists are paged. iterate() follows the pages for you:

for job in aiv.jobs.iterate(customer="cus_8421"): ...
for event in aiv.events.iterate(type="check.failed"): ...
for entry in aiv.audit_log.iterate(): ...

Errors

Every failure raises AivError with the HTTP status, a stable code (insufficient_credits, budget_exceeded, rate_limited, ...), a readable detail and the request_id to quote to support.

from aivisibilityskill import AivError
try:
    aiv.checks.create(questions)
except AivError as err:
    if err.code == "insufficient_credits": ...

GET requests and checks are retried on 429, 502, 503, 504 and network errors (twice by default, honouring Retry-After). Set max_retries and timeout in the constructor, or pass your own httpx.Client. Use it as a context manager (with Aiv(...) as aiv:) to close connections when done.

Everything else

aiv.projects (tracking, schedules, visibility, competitors, sources, facts, history, site checks, research), aiv.opportunities and aiv.fixes (what to fix and proof it worked), aiv.actions, aiv.webhooks, aiv.events, aiv.keys, aiv.account, aiv.locales, aiv.pricing() and aiv.estimate(). Responses are the API's JSON as dicts. For anything not wrapped yet: aiv.request("GET", "/v1/...").

API reference: api.aivisibilityskill.com/docs. Support: support@aivisibilityskill.com.

MIT licensed.