> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://docs.usescout.sh/llms.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs.usescout.sh/_mcp/server.

# FAQs

## About the product

**What is Scout?**
A web-intelligence API for AI agents. You send JSON, you get JSON —
ranked search results, cleaned page content, multi-step research with
citations, scheduled monitors. There is no SDK to install and no
scraping infrastructure to run.

**How is Scout different from a Google API?**
Google's APIs return raw SERP rows. Scout adds objective-aware
re-ranking, page extraction, focused excerpts, summaries, multi-step
research with planner and citations, and scheduled monitors. The
output is shaped for direct consumption by LLMs and agent frameworks.

**Where does the data come from?**
The live public web. Scout fetches every result and page at request
time. Results are not pre-indexed or cached for more than a short
freshness window, so you always see what is online right now.

**Is content cached?**
SERPs are cached briefly to absorb retries and rate-limited bursts.
Page extractions cache by URL with a configurable max age — the
default is short. LLM outputs (re-ranking, highlights, summaries) are
cached per-objective per-input so identical requests do not pay twice.

## Authentication and access

**How do I get an API key?**
Sign in at [platform.usescout.sh](https://platform.usescout.sh) and
create one under **Settings → API keys**. Keys are scoped to your team.

**Can I have multiple keys?**
Yes — one per service or environment is the recommended pattern. Each
can be rotated or revoked independently.

**What happens when I rotate a key?**
The old key keeps working for 24 hours after the new one is issued, so
you can roll your deployments without downtime.

## Limits and pricing

**How much does a call cost?**
See [Pricing](/pricing) for the full table. The short version: 1 credit
per search query, 2 per extracted URL, 1 per agent step on a task,
plus opt-in line items for re-ranking, content extraction and summaries.
Plans start at $0 (Free, 1,500 credits/mo), $29/mo (Builder, 5,500),
and \$99/mo (Scale, 25,500). Annual saves about 20%.

**What happens if I hit my monthly cap?**
On the free plan, requests return `429 Too Many Requests` until the
next reset. On Builder and Scale, overage is billed at the per-credit
rate of your plan and the API keeps serving.

**Is there a per-key rate limit?**
Not yet. Today the limits are at the account level and on a shared
concurrency pool. See [Rate limits](/rate-limits) for the practical
shape.

## Data and privacy

**Do you store the queries I send?**
Yes — query text and metadata are retained for billing, debugging,
and abuse prevention. You can purge specific records from the
dashboard. We never share queries with third parties.

**Do you store the results?**
SERPs and extracted pages are cached briefly to absorb retries; that
cache is not personalized and is shared across accounts. The
re-ranking, highlights, summary, and research outputs that were
generated for your specific request are retained on your run record
until you delete it.

**Are the URLs Scout reads recorded anywhere?**
The list of URLs visited during a research task is part of the run
payload (`sources[]` and the `basis[]` field). It is visible to anyone
with that `task_id` plus a valid API key for the same team.

**Can I use Scout for personal data?**
Only data that is public on the open web. Scout does not authenticate
into gated pages and does not work with credentials. If a request
asks Scout to read a page that requires login, the response will
contain an `errors[]` entry.

## Reliability

**What is your uptime?**
A live status page is planned. In the meantime, Scout's typical
synchronous-endpoint p95 is under 5 seconds; async runs depend on the
shape of the work. If you need an SLA, get in touch.

**What do I do when I see a 503?**
Retry with exponential backoff. See [Rate limits](/rate-limits) for
the recommended pattern. Most 503s are transient and clear within
seconds.

**Where can I report a bug or request a feature?**
Email [support@usescout.sh](mailto:support@usescout.sh). Please
include a `search_id`, `task_id`, or other run identifier when you
can — it makes debugging much faster.

## Integration

**Can I use Scout from the browser?**
You can, but you should not embed your API key in client-side code.
Call Scout from your own backend and forward results to the
browser. Same principle as any API key.

**Is there a Python or TypeScript SDK?**
Not yet. The API is small enough that a bare HTTP client (`requests`,
`fetch`, `httpx`) is usually all you need. Official SDKs are on the
roadmap.

**Does Scout work with LangChain, LlamaIndex, etc.?**
Anywhere that accepts a tool spec or an HTTP retriever, you can plug
Scout in directly. Native adapters for the popular agent frameworks
are on the roadmap.

**Can I run Scout self-hosted?**
No. Scout is offered as a hosted API only.