ask_ai_how_it_describes_my_business
ChatGPTAsk a web-search AI model three questions about your business (free, with sign-in). Reports whether each answer mentions the business and cites its website. One run per account every 30 days. Returns a job_id at once: poll get_free_tool_result with it right away until status is done. Args: site: Your site's domain, only when you manage several. Returns: The job_id and poll interval; the finished result (the questions, the answers, and mention and citation counts) comes from get_free_tool_result.
check_ai_bot_access
ChatGPTStart a free check of which AI crawlers a site serves; no sign-in needed. Needs only the site URL. Requests the site once each as GPTBot, OAI-SearchBot, ClaudeBot and PerplexityBot and reports who received a real page rather than a block or challenge. Returns a job_id at once: poll get_free_tool_result with it right away until status is done, then read result. Args: url: The site's public URL, e.g. https://example.com. email: Optional; leave it out unless the person asks Norg to follow up with them about the result. Not used for the check. Returns: The job_id and poll interval; the finished result (which crawlers were served, which were not, and robots.txt rules) comes from get_free_tool_result.
check_ai_visit_analytics_eligibility
ChatGPTCheck whether your website is served through Cloudflare (free). Changes nothing. Free AI-visit capture is available on Cloudflare, AWS CloudFront, Fastly, Bunny CDN, Shopify (Orange-to-Orange), the WordPress plugin and Vercel middleware. This is the pre-check for the Cloudflare route: it reads your site's own response headers and reports whether Cloudflare is in front of it; for the other platforms use install_on_cdn with the matching cdn. A signed-in caller with no account yet gets a free signup link instead. Args: site: Which of your sites to check, by domain. Needed only when you own or administer more than one; omit otherwise. Returns: Whether the site is on Cloudflare (true/false/unknown) with the evidence, or a note that a website address is needed first.
check_page_ai_readiness
ChatGPTStart free AI-readiness checks on one web page; no sign-in needed. Needs only the page URL. Fetches the page once and reports pass/fail on structured data, headings, title, meta description and machine-readable dates, plus the site's robots.txt AI-crawler rules, llms.txt and sitemap. Each failed check says how to fix it. Returns a job_id at once: poll get_free_tool_result with it right away until status is done, then read result. Args: url: The full public URL of the page, e.g. https://example.com/about. email: Optional; leave it out unless the person asks Norg to follow up with them about the result. Not used for the check. Returns: The job_id and poll interval; the finished result (a score, the check list with fixes, the raw measurements) comes from get_free_tool_result.
compare_ai_readiness_with_competitor
ChatGPTMeasure your site and one competitor's side by side (free, with sign-in). Schema entities, heading structure, machine-readable dates, AI-crawler rules and llms.txt on each homepage plus one product page. One comparison per account every 30 days; a further call says when the next one is available. Returns a job_id at once: poll get_free_tool_result with it right away until status is done. Args: competitor_url: The competitor's homepage, e.g. https://rival.com. site: Your site's domain, only when you manage several. Returns: The job_id and poll interval; the finished result (both measurement rows and who leads on each measure) comes from get_free_tool_result.
continue_cdn_install
ChatGPTPerform the next step of a CDN install that is now possible, and report progress. Takes no credentials. Some CDN installs cannot finish in one go: a CloudFront stack builds for several minutes before NORG can attach its router to the distribution, and Bunny or Fastly need the customer's DNS record in place before NORG can attach their hostname. This tool performs whichever of those steps is now possible on the customer's CDN, using the access the customer gave on NORG's setup page, then reports where the install stands. It removes nothing, and calling it again only repeats the check or continues the same install. Args: site: Which of your mirrors to act on, by domain. Needed only when you own or administer more than one live mirror; omit otherwise. Returns: The install's progress, the hosting state, and what to do next.
create_free_account
ChatGPTFinish creating the person's free Norg account (free, with sign-in). Use when a signed-in tool replied state no_account: the person has signed in but has no account yet. Ask them for the business name and website, then call this. If their email is not verified yet it replies email_unverified and re-sends the link; call again with the same details once they have clicked it. It creates the free account at once — no other step and no web page needed. Calling it again for someone who already has an account changes nothing. Set email_updates only if the person asks for the monthly AI traffic report and product emails; it defaults to off. Args: business_name: The business's name, e.g. "Selleys". website: The business's homepage, e.g. https://www.selleys.com.au. email_updates: True only if the person opted in to the monthly AI traffic report and occasional product emails. Returns: state ready (with setup_link for free AI traffic analytics), already_has_account, or profile_unsaved — or an error: email_unverified, website_taken, an invalid business name or website, or not_signed_in.
delete_ai_visit_analytics
ChatGPTPermanently delete your recorded AI-visit analytics (free, with sign-in). Covers AI-agent visits (and their summaries) and AI-referral page-views. Call first without confirm to see what would be deleted, ask the person, and only on their yes call again with confirm=true. New visits are still recorded while an install is live; uninstall_from_cdn stops collection. Args: confirm: True only after the person confirmed the deletion. site: Your site's domain, only when you manage several. Returns: Row counts per table and confirmation_required or deleted.
generate_llms_txt
ChatGPTBuild an llms.txt for your site from your audit's pages (free, with sign-in). Needs a finished AI Search Audit started with this account's email. Args: site: Your site's domain, only when you manage several. Returns: The llms.txt text, the page count, and where to save it.
generate_page_jsonld
ChatGPTWrite schema.org JSON-LD for one of your pages (free, with sign-in). Five pages per account every 30 days. Returns a job_id at once: poll get_free_tool_result with it right away until status is done. Args: url: The page to describe. kind: "WebPage" (default), "Organization" or "LocalBusiness". site: Your site's domain, only when you manage several. Returns: The job_id and poll interval; the finished result (the JSON-LD, a script tag to paste into the page head, and steps) comes from get_free_tool_result.
generate_robots_txt_ai_rules
ChatGPTShow a site's robots.txt AI-crawler rules and write a new stanza. No sign-in needed; needs only the site URL. Lists the AI crawlers robots.txt blocks today and returns a stanza that allows (default) or blocks the main AI crawlers. Args: url: The site's public URL, e.g. https://example.com. policy: "allow" (default) or "block". email: Optional; leave it out unless the person asks Norg to follow up with them about the result. Not used for the rules. Returns: The current rules, the stanza to add, and how to install it.
get_ai_search_audit_report
ChatGPTRead your completed AI Search Audit report in full (free, with sign-in). Use it to explain findings and decide what to fix first. Args: site: Your site's domain, only when you manage several. Returns: Score, band, page counts, the report link and the full findings.
get_ai_search_audit_status
ChatGPTCheck how far an AI Search Audit has got, by its id. Anonymous and changes nothing. Returns only the lifecycle state and progress — never the scored findings, which are delivered by email to the address that started the audit. Args: audit_id: The id returned by start_ai_search_audit. Returns: The status, a progress fraction and a short progress message (a held audit adds held_reason and a detail to tell the person), or a not-found result for an unknown id.
get_ai_visit_analytics_status
ChatGPTShow whether free AI-visit capture is running on your site, and the numbers. Changes nothing. Reports one capture state (not installed, awaiting the first heartbeat, capturing, or not capturing) plus the AI-visit counts behind it over the last 24 hours and the reasons when it is not capturing. Free AI-visit capture is available on Cloudflare, AWS CloudFront, Fastly, Bunny CDN, Shopify (Orange-to-Orange), the WordPress plugin and Vercel middleware. A signed-in caller with no account yet gets a free signup link instead. Args: site: Which of your sites to check, by domain. Needed only when you own or administer more than one; omit otherwise. Returns: The capture state, the site, the worker version, the AI-visit counts for the last 24 hours, and any reasons capture is not running.
get_business_profile
ChatGPTShow what Norg holds about your business and what the audit found (free, with sign-in). Args: site: Your site's domain, only when you manage several. Returns: The profile, audit facts, and the fields update_business_details edits.
get_crawler_analytics
ChatGPTShow which AI crawlers visited in the last 30 days. Changes nothing. Over a fixed 30-day window: visit totals by crawler company, visits per bot split into verified and unverified, and the latest visits, which include ones too recent for the company totals. Args: site: Which of your mirrors to act on, by domain. Needed only when you own or administer more than one live mirror; omit otherwise. Returns: The window length and a note that every figure covers only the last 30 days, visit counts per company, verified and unverified visits per bot, the latest visits, and a note on what the figures cover.
get_free_tool_result
ChatGPTRead the result of a free tool that returned a job_id (free). Changes nothing. check_page_ai_readiness, check_ai_bot_access, compare_ai_readiness_with_competitor, generate_page_jsonld and ask_ai_how_it_describes_my_business queue their work and return a job_id; call this right away until status is done or failed. Each call waits up to 20 seconds for the job, so call it again whenever it answers queued or running instead of stopping. Works without sign-in for the anonymous checks; a signed-in tool's job is visible only to the account that started it. Args: job_id: The job_id the free tool returned. Returns: status queued, running, done (with result) or failed (with the reason; a failed run's allowance is given back).
get_job_queue
ChatGPTShow how busy the crawl and knowledge-base build queues are. Anonymous, read-only, and safe before any account exists. Returns, per queue, the pending and processing counts, the oldest/newest pending times, and recent throughput — enough to estimate a wait. No job, workspace, user id, URL or domain is returned. Args: include_pending_times: Also return each queue's pending-job times (oldest first). Off by default to keep the reply short. Returns: Per-queue counts, timestamps and throughput.
get_mirror_content
ChatGPTShow the caller's mirrored HOMEPAGE in its three sections. Changes nothing. Homepage only: the crawled text, the research text NORG wrote, and the question-and-answer block. Read it before editing so the customer can see what they are about to replace. Args: body_chars: Optional. Return only the first N characters of the two text sections. Omit for the full text — always read the full research text before editing. site: Which of your mirrors to act on, by domain. Needed only when you own or administer more than one live mirror; omit otherwise. Returns: The three sections, whether the page is live, and the research text's saved versions.
get_mirror_status
ChatGPTShow the current state of the caller's mirror and the next step. Changes nothing. Reports where the mirror is in its lifecycle, the published URL once live, and what to do next. Args: site: Which of your mirrors to act on, by domain. Needed only when you own or administer more than one live mirror; omit otherwise. Returns: The mirror state, the next step, and the served URL when active.
get_mirrored_pages
ChatGPTList every page in the caller's mirror: its URL, crawl date and length. Changes nothing. Covers all mirrored pages as crawled-text metadata; the homepage's generated research and Q&A are get_mirror_content's job. Args: site: Which of your mirrors to act on, by domain. Needed only when you own or administer more than one live mirror; omit otherwise. Returns: The page count and, for each page, its URL, a preview URL, render status, last-crawled date and word count.
get_service_overview
ChatGPTDescribe what AI Retrievability by Norg does and list this connector's tools. Anonymous, takes no arguments and changes nothing. Describes the copy of a website that NORG builds and serves to AI agents (the mirror), what it covers, the two hosting options, and the CDNs and platforms NORG can set it up on. Returns: The service description, the hosting options, the supported CDNs and platforms, help and dashboard links, and `call_order`: the name of every tool on this connector, in the order they are usually used.
install_on_cdn
ChatGPTStart installing NORG on the customer's CDN via NORG's secure setup page. Takes no credentials. Returns a signed link, valid for one hour, to NORG's setup page for the chosen platform, where the customer enters their own CDN credential and completes the install; the call itself makes no call to the CDN. For WordPress and Vercel the link is the NORG dashboard page where the plugin or middleware keys are issued. After the page is submitted, continue_cdn_install reports progress. A signed-in caller with no NORG account yet gets a signup link instead. Args: site: Which of your sites to set up, by domain. Needed only when you own or administer more than one; omit otherwise. cdn: The platform that serves the site; defaults to Cloudflare. Returns: The signed setup link (or, for Vercel, the NORG dashboard link), the next step, and for a CDN the names of the fields the setup page will ask for — or a signup link when you have no account yet.
measure_javascript_hidden_content
ChatGPTMeasure how much of a page AI crawlers never receive (sign-in). Fetches the initial HTML and the fully rendered page, strips both to text, and reports raw tokens, rendered tokens, the hidden tokens produced only by JavaScript and the hidden share, plus which of title, meta description, canonical link, JSON-LD and h1 appear only after rendering. One run per account every 30 days. Args: url: The full public URL of the page to measure. site: Your site's domain, only when you manage several. Returns: The token counts, the hidden share, and an advice block naming the rendered-only metadata and how a NORG mirror closes the gap.
move_mirror_to_norg_hosting
ChatGPTMove a self-hosted mirror back to NORG hosting. Removes NORG's worker and its routes from the customer's CDN account, using the access the customer already gave on NORG's setup page (or turns the worker off where it cannot be removed), then serves the mirror from a NORG sub-domain. The customer's own website is not otherwise changed. Takes no credentials. Calling it again once the mirror is NORG-hosted changes nothing. Args: site: Which of your mirrors to act on, by domain. Needed only when you own or administer more than one live mirror; omit otherwise. Returns: The new hosting state, the mirror URL and what was removed from the customer's CDN.
my_account
ChatGPTShow the caller's NORG account: business name, service, hosting and status. Read-only. Reports the business name on the account, which NORG service it has and what that service covers, how the mirror is hosted, and whether the mirror is currently being served. Args: site: Which of your mirrors to act on, by domain. Needed only when you own or administer more than one live mirror; omit otherwise. Returns: The business name, the service and what it covers, the hosting type, the active/not-active status, and a link to the NORG dashboard where account settings are managed.
run_brand_gap_analysis
ChatGPTCompare how AI describes your business against competitors (free, with sign-in). Asks a web-search-enabled AI model fixed questions about the business and each competitor (what they offer, who it is for, what reviews cite), and measures every site with the audit benchmark (schema.org types, llms.txt, machine-readable dates, AI-crawler rules). Returns AI-claim gaps (proof points a competitor's description earns that yours does not) and structural gaps, each naming the free tool that fixes it. Needs a homepage on the profile and a verified email. A free account includes ONE analysis for the life of the account (one competitor, one question per site); a second request returns account_type_required — tell the person more brand gap analyses need a different account type. Other accounts get a set number every 30 days (a further call returns allowance_exhausted with the reset date) or the full analysis with up to three competitors. Returns a job_id at once: poll get_free_tool_result with it right away until status is done. A run that fails is not counted. Args: competitor_urls: One to three competitor homepage URLs, e.g. ["https://rival.com"]. At least one is required; extra URLs beyond the account's limit are ignored. Your own site is refused (same_site). site: Your site's domain, only when you manage several. Returns: The job_id and poll interval, or an error: profile_incomplete, competitors_required, invalid_primary_url, same_site, account_type_required, allowance_exhausted (with resets_at) or email_unverified.
send_feedback
ChatGPTSend the customer's feedback, question, or support request to NORG. Use it when the customer wants to reach the NORG team from inside the conversation. The message is emailed to the team. Args: message: What the customer wants to tell NORG. Required, non-empty, up to 5000 characters. category: One of "feedback", "question" or "support" (default "feedback"). contact_email: Optional reply-to address; only included in the email. contact_name: Optional name so the team knows who wrote. site: Which of your mirrors to act on, by domain. Needed only when you own or administer more than one live mirror; omit otherwise. Returns: A confirmation, or a structured error explaining what to fix (e.g. an empty or over-length message).
set_email_tracking
ChatGPTTurn open and click tracking in NORG's emails off or on (free, with sign-in). With tracking off, emails carry no tracking pixel or tracked links, and opens or clicks on earlier emails are no longer recorded. Args: enabled: False to turn tracking off (default), True to turn it on. site: Your site's domain, only when you manage several. Returns: The new setting.
set_weekly_ai_visibility_email
ChatGPTTurn a weekly email re-checking your homepage on or off (free, with sign-in). Args: enabled: True to start the weekly email (default), False to stop. site: Your site's domain, only when you manage several. Returns: The new setting.
start_ai_search_audit
ChatGPTStart a free AI Search Audit of a website and email the report. Anonymous: it needs only the site URL and an email address, and no sign-in. It audits how well AI agents can read the site and emails the report to the address given. If a recent report for the same site exists it is re-sent rather than re-run. A held request comes back as status held with held_reason and a detail to tell the person: why it waits and how to run it straight away (e.g. an email address at the site's own domain). Args: url: The public website URL to audit, e.g. https://example.com email: The address the finished report is emailed to. Returns: The audit id and status, whether a recent report was re-sent, a poll interval, and a note that the report is emailed to the address given — or a plain-language reason it could not start.
trigger_recrawl
ChatGPTRe-crawl the caller's site now and refresh the mirror. For when the site changed and the customer does not want to wait for the next scheduled crawl. Rate-limited to once every 7 days per site; a call inside that window is refused with the days remaining. Args: site: Which of your mirrors to recrawl, by domain. Needed only when you own or administer more than one live mirror; omit otherwise. Returns: Confirmation the recrawl was queued and when the next manual recrawl is available, or the cooldown message.
uninstall_from_cdn
ChatGPTRemove NORG's worker and its routes from the customer's own CDN account. Takes no credentials. Afterwards AI agents get the customer's ordinary pages and the mirror is not served anywhere; the customer's own website is not otherwise changed. For CloudFront, Bunny and Fastly, NORG removes the worker using the access the customer already gave on NORG's setup page. For Cloudflare, or when no access is stored, it returns a signed one-hour link to that setup page, where the customer confirms the removal with their own credential. Args: site: Which of your sites to act on, by domain. Needed only when you own or administer more than one; omit otherwise. Returns: What was removed and what to do next, or the signed setup-page link to finish the removal there.
update_business_details
ChatGPTChange the business name and crawled-text setting NORG stores for a mirror. Edits two settings in the customer's NORG account and nothing else: the business name, and whether the mirror includes the text crawled from the customer's own pages (include_existing_page_text). It only updates data NORG stores: it does not contact the customer's website, their CDN or any other service, and it does not republish the mirror; the mirror picks the change up at its next refresh (the scheduled crawl, or trigger_recrawl). Send only the field(s) to change. The website address cannot be changed here. Changes show up in get_mirror_status and my_account. Args: business_name: The business's name, as it should appear. Omit to leave it unchanged. include_existing_page_text: true or false — whether the mirror also publishes the text crawled from the customer's own pages. Omit to leave it unchanged. site: Which of your mirrors to edit, by domain. Needed only when you own or administer more than one live mirror; omit otherwise. Returns: Which fields changed and their new values, or a reason nothing changed. At least one field is required.
update_mirror_content
ChatGPTReplace the research text and/or the Q&A on the mirrored HOMEPAGE. Homepage only, and each section is replaced whole, never merged. The crawled text is never touched. The edited page is republished at once to the public mirror that AI agents read, served from NORG's CDN storage, replacing text that is already live; the research text keeps recent versions. Show the customer the current sections first. Args: body: The complete new research text, in Markdown. Leave empty to keep the current research text. qa: The complete new Q&A as a list of {"question","answer"} objects. Omit to keep the current Q&A. title: Optional new page title. Left empty, the title is kept. site: Which of your mirrors to edit, by domain. Needed only when you own or administer more than one live mirror; omit otherwise. Returns: Which sections changed and the mirror URL, or a reason nothing changed. At least one of title, body or qa is required; a title-only edit is fine.