AI tools reference
The exact capabilities the SiteOps assistant can use on your behalf. Every one runs on the backend, scoped to your account. You never call these directly - you ask in plain language and the AI picks the right one - but here's the full spec.
Monitoring - read (no changes, run instantly)
| Tool | What it returns | Ask it like |
|---|---|---|
list_monitored_sites |
Your sites: URL, active/paused, frequency, site health scan mode | "what am I monitoring?" |
get_site_monitoring |
Your config: frequency, alert thresholds, site health scan, business hours, timezone, plan | "what are my settings?" |
get_site_health |
Open incidents + per-site scheduling/last-run status | "how's my site doing?" |
check_site_ssl |
SSL certificate + domain-expiry (days left, issuer, dates, warnings) | "when does my SSL expire?" |
get_monitoring_incident |
Full detail on one incident by id | "explain incident #123" |
Monitoring - actions & changes
| Tool | What it does | Safety |
|---|---|---|
run_site_check |
Queues a real on-demand browser scan of your site now | Rate-limited per user (short cooldown); counts toward monitoring usage |
setup_site_monitoring |
Starts monitoring a new URL with an AI-drafted config | Preview → confirm before it creates anything |
configure_site_monitoring |
Changes monitoring settings (frequency, thresholds, site health scan, pause, timezone…) | Preview → confirm; values are clamped to your plan |
Journeys
Journeys are the core of SiteOps - real multi-step flows (checkout, login, signup, booking, contact form) replayed in a real browser on a schedule. You drive them in plain language and the AI runs the journey builder and monitoring for you. Full guide: User journeys.
| Ask the AI to | What happens | Safety |
|---|---|---|
| Build a journey ("build a journey for example.com that adds a product to the cart and checks out") | The AI opens the site, reads the real buttons and fields, and assembles the steps from what actually exists, then saves and monitors it | Builds from real controls only (it cannot invent steps); a build that ends on a real form submit asks yes/no first |
| Suggest journeys ("suggest journeys for my site") | Crawls the site and returns the 3-4 flows that matter most as plain-English prompts, ready to build in one click | Read-only crawl |
| Set a journey's run frequency ("set my checkout journey to run every 15 minutes") | Changes how often the journey replays | Preview -> confirm; clamped to your plan's floor |
| Explain or fix a journey incident ("explain the incident on my checkout journey") | Explains the exact failing step in plain English; for a renamed/moved control it can apply a one-click fix grounded in the live page and re-run to confirm | Explain is read-only; a fix is previewed and re-run before it resolves |
Research & web
| Tool | What it does |
|---|---|
web_search |
Searches the open web and returns a summarized answer with sources |
browse_page |
Loads a URL in a real headless browser on the server and returns the JS-rendered title, text, and console errors (optionally a screenshot) |
Foundations (available in any chat)
| Tool | What it does |
|---|---|
manage_memory |
Remembers and recalls facts across your conversations |
| Codebase tools | If you attach a repo, the AI can read and (with confirmation) fix code - read_file, search, grep, edits, etc. |
How permissions work
- Read tools run immediately and can't change anything.
- Change tools (
setup_site_monitoring,configure_site_monitoring) always preview first and wait for your explicit "yes." - Everything is scoped to your account - the AI authenticates as you and can't see or touch other users' data. If your session isn't signed in, monitoring tools fail closed.
Output limits (so the AI stays fast, not overwhelmed)
Reads and searches are capped at production-sane sizes and paginate cleanly:
- File reads: up to 1000 lines per call (large files return a preview + "read a specific range").
- Search / grep: up to 50 matches per call.
- Directory listings: up to 300 entries.
These keep responses focused instead of dumping huge blobs - the AI is told exactly how to fetch more when it needs it.
Want your own agent (Claude Code, Cursor, Codex, a custom bot) to understand and drive SiteOps? See For AI agents.