FAQ
Common questions about ingestion, blocking, pricing, and what happens if you don't set things up.
What happens if I don't set up forwarding?
AI crawlers continue to hit your canonical pages as before. You lose:
- Per-article policy enforcement (you can't block / paywall them on the
way in) - Per-agent traffic analytics
- The agent-payment revenue path
Your MCP endpoint at oasy.<your-domain>/mcp still works for clients
that explicitly opt in (Claude Desktop, ChatGPT connectors, Cursor) —
that's a different traffic path entirely.
What happens if I don't verify my domain?
Your MCP works at the long form:
https://api.oasy.ai/api/mcp/<orgId>.
You can still hand that URL to Claude / ChatGPT / Cursor. You just don't
get the clean oasy.<your-domain>/mcp vanity URL, and the AI-traffic
landing pages don't have a host to live on.
Can I block a specific agent?
Not yet at the agent level (per-agent rules are on the roadmap). You can
block specific articles via the Block policy, and the content
rules engine gives you path patterns (/premium/*), embargo windows
("articles younger than 48h"), territorial restrictions, and
time-window pricing — all enforced at serve time on every surface.
How often do you re-crawl my site?
Your ingest sources (sitemaps, RSS/Atom feeds) are re-polled
automatically on an hourly sync, and the dashboard's Ingestion panel has
a one-click Sync sources + drain plus CSV/sitemap upload and
single-URL ingest. Agents asking for a URL we haven't crawled trigger an
on-demand ingest of that page (HTML and PDF).
What if I take down an article?
It stays in our index until you delete it via support or the next
re-ingest. The body that gets served is the version we crawled. Article
takedowns + corrections are the most common roadmap ask — coming soon.
Can I see who's reading what?
Yes — every MCP call is in the Recent Calls table on /mcp, with
the agent UA, tool, status, and latency. With OAuth on, each row also
carries the install's client_id and (if captured) email.
How do you handle robots.txt?
Our own ingest crawler respects robots.txt. The forwarding rule is
different — it lives on your CDN and runs before
robots.txt is consulted. If you want to keep your robots.txt as-is for
agents that respect it, that's fine: those agents will skip your site
entirely. The forwarding rule only catches the ones who don't.
Is there an SLA?
Service levels are measured and published live at
/status: uptime from 5-minute synthetic probes, latency
percentiles per surface, and error rates, scored against 99.9%
availability / search p95 < 300 ms / error rate < 1%. Contractual SLAs
with credits come with enterprise agreements — the measured numbers are
what we sign against.
Can I use Oasy with a non-news publication?
Absolutely. Reviews sites, documentation hubs, data portals, recipe
sites — anything with discrete pages an agent might want to read.
The four built-in tools (search_articles, get_article, etc.) are
generic; the "article" framing is just naming.
Refund policy on agent payments?
Agent payments are non-refundable once the article body is served — same
posture as any paid API call. If an article serves but is malformed /
404s upstream, ping support and we'll credit it.
Updated 7 days ago