A Clicky MCP built out of questions, not query parameters
18 nodes covering visitor and action totals, engagement and time, top pages, entry and exit pages, referrers, searches, locations, browsers, devices, screen sizes, downloads and events, goals and campaigns, segmentation, the session log, uptime and alerts. Every node reads and nothing writes, because Clicky's write surface is a separate key this connector deliberately does not hold.
What the Clicky MCP connector does
The Clicky connector for Mattering Automations is 18 nodes on Clicky's stats API. All 18 are reads, none is destructive, and each one names a question a person actually asks: which pages were read, where the traffic came from, what people searched for, how long they stayed, who is on the site right now, which goals converted, whether the site was up. There is also a catch-all node for anything the typed nodes group differently from the way you need it.
Every capability is a node on the Automations canvas, and every node speaks MCP, the open protocol for connecting AI assistants to tools. Claude, ChatGPT, and any MCP-capable assistant can use them, alongside Matter, the AI assistant built into the platform. So you can ask an assistant which blog posts lost traffic last month and have it read the answer out of Clicky, and the same nodes then run on a schedule with nobody present, which is the part a local assistant config cannot do.
The shape of this connector is a direct consequence of the shape of Clicky's API, and it is worth understanding before you plan anything. Clicky is not a resource API with an endpoint per thing. It is essentially one address, the stats endpoint, whose meaning is chosen by a type parameter, and Clicky documents 57 values for it. Two obvious ways to model that were both rejected. One node per type would be 57 near-identical nodes differing by a single string, which is a scrolling list rather than a palette. One node with a free-text type box would be the generic HTTP node wearing a Clicky costume: nothing to discover, no per-report sample output, and every typo surfacing as an empty report rather than as an error.
So each node owns a report family and offers only the type values that belong to it, validated before the request goes out. Ask Clicky: Top Pages for a visitor tally and it returns an error naming the values it accepts, instead of returning nothing and leaving you to guess. The grouping is a partition rather than a selection: all 57 documented values belong to exactly one node, so nothing in Clicky's reporting surface is unreachable, and Clicky: Run Any Report Type accepts any of them for the cases where our grouping is not your grouping.
Authentication is two values, and getting one of them wrong is the most likely thing to go wrong on your first run. Clicky identifies a request with a numeric Site ID and a sitekey, both sent as URL query parameters; it documents no header form, no bearer token and no signature. Clicky describes the sitekey as a 12 to 16 character string assigned when you first register a site, and it is the READ key. Clicky mints a separate admin sitekey, which its manual tracking documentation describes as the key that allows you to write data to Clicky, with the instruction not to share it with anyone else. Pasting that one here is the classic first failure, because it lives on the same site preferences page. This connector only reads, so it never needs the admin key and there is nowhere to put it.
Because the credential rides in the query string rather than in a header, it is written down by more things than a header secret is: proxy and CDN access logs, browser history, the vendor's own request logs, and any error string that echoes a URL back. That is a property of Clicky's API and no connector can undo it. What this one does do is register the sitekey with the per-execution redaction context so it is scrubbed from node outputs and execution records, pass it to the HTTP client as a separate parameter mapping rather than formatting it into a URL, and scrub it out of transport error text before returning, because an ordinary timeout would otherwise print the key into your workflow output on the way past.
API access itself is not plan-gated: Clicky lists it among the basics that every plan gets, the free one included. What your plan changes is what comes back, and that is the single most useful thing to know before you build a dashboard on this. Clicky's API documentation states that free accounts are limited to the last 30 days of history and 100 results per request, and that the maximum date range is 366 days for Pro users and 31 days for everyone else. Its pricing page puts the same boundary on retention: free plans keep 30 days of history, paid plans keep it indefinitely, with individual visitors and actions guaranteed for six months because of the space they take. Clicky enforces the caps by truncating rather than by erroring, so a workflow asking for a quarter on a free account looks perfectly healthy and quietly returns a month. Two report families are gated by feature rather than by cap: goals and revenue are premium features that the free plan does not include, and uptime monitoring is listed as Pro Plus and above, so base Pro does not have it either. Neither Clicky: Goals and Campaigns nor Clicky: Uptime Checks has anything to return without them, so treat an empty result there as its own case rather than as a zero. Test Connection cannot warn you about any of this, because a free credential is a perfectly valid credential; the success message says so in as many words.
One credential means one site. A Clicky sitekey is minted for a single site and is meaningless against another site's id, which is how the Clicky dashboard already frames it, so reporting on a second site means storing a second credential rather than passing a site id into a node. There is no site picker for the same reason: the one Clicky endpoint that could list your sites authenticates with your account username and password instead of a sitekey, and we are not asking anyone to put an account password into a read-only reporting credential.
Two things worth knowing about the output before you wire anything to it. Clicky answers with a nested structure that puts the metric's identity and the row's date on the row's ancestors rather than in the row itself, so a chart bound naively to the items cannot tell a page view from a bounce rate or Monday from Tuesday. Every node therefore returns data.reports, which is Clicky's response verbatim, alongside data.rows, a flat view this connector derives where the type and the date are real columns. Bind charts, databases and Analytics datasets to data.rows. And a Clicky 200 is not automatically a success: Clicky can answer HTTP 200 with an error payload in the body, so the connector inspects the body as well as the status and fails the node with the vendor's own message rather than presenting an auth failure as a quiet empty report.
One connection, three surfaces
- Visual automations. Drag Clicky nodes onto the canvas and wire them to triggers, conditions, and other apps.
- AI agents, yours or ours. These nodes speak MCP, the open protocol, so Claude, ChatGPT, and any MCP-capable assistant can use them, and Matter, the agent built into the platform, uses them natively.
- The whole platform. Combine with native tasks, projects, and other connectors, no middleware in between.
Every Clicky node, ready to drop on the canvas
18 nodes, generated from the connector itself so this list always matches what ships. Expand a category to browse.
Read and query nodes 18
Clicky: Action Totals
How much happened: total actions, page views, file downloads, outbound link clicks and javascript click events. These are the numbers you CAN add up. Every type this node offers is an event count rather than a deduplicated or averaged figure, so a fortnight really is the sum of its fourteen days, and a dashboard may roll daily partitions into weeks and months freely. The averages that look like they belong here (actions-average, time-average, bounce-rate) are deliberately on the Clicky: Engagement and Time Totals node instead, because they are rates and must never be summed. Output shape: data.reports is Clicky's own response verbatim, one entry per requested type, each holding dates and then items. data.rows is a flat view DERIVED by this connector, where every item carries its own type and date as real columns, because Clicky puts both on the row's parents rather than in the row. Bind charts and datasets to data.rows. Plan caps, from Clicky's own docs: a FREE account can only reach the last 30 days of history and returns at most 100 results per request, so a longer window or a bigger limit quietly returns less than you asked for. The date range itself is capped at 366 days on Pro and 31 days on every other plan.
Clicky: Engagement and Time Totals
How engaged people were: bounce rate, average actions per visitor, average and total time on site (in seconds and in Clicky's pretty form), plus the two engagement breakdowns that bucket visitors by how many actions they performed and how long they stayed. Read the numbers, do not add them. Every figure here except time-total is an AVERAGE, a PERCENTAGE or a formatted string, and none of those may be summed across days or averaged unweighted; Clicky returns no visitor count beside them to weight by, so the only correct move is to re-ask Clicky for the period you actually want. time-total is additive on its own but shares this node with the rates, so this connector declares nothing aggregatable here at all. Output shape: data.reports is Clicky's own response verbatim, one entry per requested type, each holding dates and then items. data.rows is a flat view DERIVED by this connector, where every item carries its own type and date as real columns, because Clicky puts both on the row's parents rather than in the row. Bind charts and datasets to data.rows. Plan caps, from Clicky's own docs: a FREE account can only reach the last 30 days of history and returns at most 100 results per request, so a longer window or a bigger limit quietly returns less than you asked for. The date range itself is capped at 366 days on Pro and 31 days on every other plan.
Clicky: Entry and Exit Pages
Where visits start and where they end: landing pages (pages-entrance) and the last page someone saw before leaving (pages-exit). Requesting both in one call is the usual pattern, since the interesting answer is usually the difference between them. Rows come back ordered by value descending and cut off at limit, so a result is a TOP N rather than the whole picture: raise limit or page through before treating it as complete, and never add two periods' top lists together, because the top 20 of January plus the top 20 of February is not the top 20 of both. Output shape: data.reports is Clicky's own response verbatim, one entry per requested type, each holding dates and then items. data.rows is a flat view DERIVED by this connector, where every item carries its own type and date as real columns, because Clicky puts both on the row's parents rather than in the row. Bind charts and datasets to data.rows. Plan caps, from Clicky's own docs: a FREE account can only reach the last 30 days of history and returns at most 100 results per request, so a longer window or a bigger limit quietly returns less than you asked for. The date range itself is capped at 366 days on Pro and 31 days on every other plan.
Clicky: Get Alert
Fetch the detail behind one Clicky alert. Clicky delivers alerts by email; each email carries an alert ID, and this node turns that ID back into structured data so an automation can act on it rather than parse an email. This is the only mode of the stats endpoint that takes no report type and no date: it wants the alert ID alone. Output shape: data.reports is Clicky's own response verbatim, one entry per requested type, each holding dates and then items. data.rows is a flat view DERIVED by this connector, where every item carries its own type and date as real columns, because Clicky puts both on the row's parents rather than in the row. Bind charts and datasets to data.rows.
Clicky: Goals and Campaigns
Everything you configured Clicky to watch: goal completions (which also carry incompleted, conversion, revenue and cost alongside the usual value), campaign performance, split-test results, and the custom data your own tracking code logged against sessions. Goals are a Pro feature on Clicky, so a free or basic account gets an empty goals report rather than an error. Revenue figures are additive across days; conversion rate is not, and this connector declares neither, so treat any rate you see as read-only. Rows come back ordered by value descending and cut off at limit, so a result is a TOP N rather than the whole picture: raise limit or page through before treating it as complete, and never add two periods' top lists together, because the top 20 of January plus the top 20 of February is not the top 20 of both. Output shape: data.reports is Clicky's own response verbatim, one entry per requested type, each holding dates and then items. data.rows is a flat view DERIVED by this connector, where every item carries its own type and date as real columns, because Clicky puts both on the row's parents rather than in the row. Bind charts and datasets to data.rows. Plan caps, from Clicky's own docs: a FREE account can only reach the last 30 days of history and returns at most 100 results per request, so a longer window or a bigger limit quietly returns less than you asked for. The date range itself is capped at 366 days on Pro and 31 days on every other plan.
Clicky: Operating Systems and Devices
What your visitors ran: operating systems, and the hardware report which names the mobile and tablet devices Clicky could identify. Desktop traffic largely does not appear in the hardware report, so read the two together rather than treating hardware as a device split of all traffic. Rows come back ordered by value descending and cut off at limit, so a result is a TOP N rather than the whole picture: raise limit or page through before treating it as complete, and never add two periods' top lists together, because the top 20 of January plus the top 20 of February is not the top 20 of both. Output shape: data.reports is Clicky's own response verbatim, one entry per requested type, each holding dates and then items. data.rows is a flat view DERIVED by this connector, where every item carries its own type and date as real columns, because Clicky puts both on the row's parents rather than in the row. Bind charts and datasets to data.rows. Plan caps, from Clicky's own docs: a FREE account can only reach the last 30 days of history and returns at most 100 results per request, so a longer window or a bigger limit quietly returns less than you asked for. The date range itself is capped at 366 days on Pro and 31 days on every other plan.
Clicky: Outbound Links and Downloads
What visitors clicked, took and watched: outbound links they followed off the site, files they downloaded, javascript events you are tracking, and video plays. All four are counted as actions, so their totals reconcile with the action tallies rather than with visitor counts. Turn Video Meta on to get extra video measurements such as average watch time alongside the play count. Rows come back ordered by value descending and cut off at limit, so a result is a TOP N rather than the whole picture: raise limit or page through before treating it as complete, and never add two periods' top lists together, because the top 20 of January plus the top 20 of February is not the top 20 of both. Output shape: data.reports is Clicky's own response verbatim, one entry per requested type, each holding dates and then items. data.rows is a flat view DERIVED by this connector, where every item carries its own type and date as real columns, because Clicky puts both on the row's parents rather than in the row. Bind charts and datasets to data.rows. Plan caps, from Clicky's own docs: a FREE account can only reach the last 30 days of history and returns at most 100 results per request, so a longer window or a bigger limit quietly returns less than you asked for. The date range itself is capped at 366 days on Pro and 31 days on every other plan.
Clicky: Referrers and Traffic Sources
Where your traffic came from. traffic-sources is the high-level split (direct, search, links, campaigns), links is every incoming referrer as a full URL, links-domains rolls those up by domain, and the two chronological variants list referrers newest first (links-recent) or only the ones seen for the very first time (links-unique), which is the one to watch for new coverage. Rows come back ordered by value descending and cut off at limit, so a result is a TOP N rather than the whole picture: raise limit or page through before treating it as complete, and never add two periods' top lists together, because the top 20 of January plus the top 20 of February is not the top 20 of both. Output shape: data.reports is Clicky's own response verbatim, one entry per requested type, each holding dates and then items. data.rows is a flat view DERIVED by this connector, where every item carries its own type and date as real columns, because Clicky puts both on the row's parents rather than in the row. Bind charts and datasets to data.rows. Plan caps, from Clicky's own docs: a FREE account can only reach the last 30 days of history and returns at most 100 results per request, so a longer window or a bigger limit quietly returns less than you asked for. The date range itself is capped at 366 days on Pro and 31 days on every other plan.
Clicky: Run Any Report Type
Run any Clicky report by naming its type, for the reports the typed nodes above do not cover the way you need. The full vocabulary, which is Clicky's entire reporting surface, is these 57 values. Popular items: custom, pages, pages-entrance, pages-exit, downloads, events, video, site-domains, links, links-domains, links-outbound, searches, searches-keywords, searches-engines, searches-rankings, searches-local, goals, split-tests, campaigns, countries, cities, regions, languages, web-browsers, operating-systems, screen-resolutions, hardware, hostnames, organizations, engagement-actions, engagement-times, segmentation, visitors-most-active, traffic-sources. Chronological items: visitors-list, actions-list, searches-recent, searches-unique, links-recent, links-unique, uptime, uptime-list. Tallies: visitors, visitors-online, visitors-unique, visitors-new, actions, actions-pageviews, actions-downloads, actions-outbounds, actions-clicks, actions-average, time-average, time-average-pretty, time-total, time-total-pretty, bounce-rate. Prefer a typed node when one fits: they carry the right defaults and the right aggregation rules, and this node deliberately declares NO measures, because what may be summed depends entirely on which type you chose. Rows come back ordered by value descending and cut off at limit, so a result is a TOP N rather than the whole picture: raise limit or page through before treating it as complete, and never add two periods' top lists together, because the top 20 of January plus the top 20 of February is not the top 20 of both. Output shape: data.reports is Clicky's own response verbatim, one entry per requested type, each holding dates and then items. data.rows is a flat view DERIVED by this connector, where every item carries its own type and date as real columns, because Clicky puts both on the row's parents rather than in the row. Bind charts and datasets to data.rows. Plan caps, from Clicky's own docs: a FREE account can only reach the last 30 days of history and returns at most 100 results per request, so a longer window or a bigger limit quietly returns less than you asked for. The date range itself is capped at 366 days on Pro and 31 days on every other plan.
Clicky: Screen Resolutions
Screen resolutions your visitors browsed at, ranked by visits. The practical use is deciding which breakpoints a redesign has to hold up at, so pair it with Clicky: Operating Systems and Devices rather than reading it alone. Rows come back ordered by value descending and cut off at limit, so a result is a TOP N rather than the whole picture: raise limit or page through before treating it as complete, and never add two periods' top lists together, because the top 20 of January plus the top 20 of February is not the top 20 of both. Output shape: data.reports is Clicky's own response verbatim, one entry per requested type, each holding dates and then items. data.rows is a flat view DERIVED by this connector, where every item carries its own type and date as real columns, because Clicky puts both on the row's parents rather than in the row. Bind charts and datasets to data.rows. Plan caps, from Clicky's own docs: a FREE account can only reach the last 30 days of history and returns at most 100 results per request, so a longer window or a bigger limit quietly returns less than you asked for. The date range itself is capped at 366 days on Pro and 31 days on every other plan.
Clicky: Searches and Keywords
Everything Clicky knows about search: full search queries (searches), the same queries split into individual keywords (searches-keywords), which engines sent the traffic (searches-engines), your Google rankings ordered best to worst (searches-rankings), local-search terms (searches-local), and the chronological views of queries seen most recently (searches-recent) or for the first time ever (searches-unique). Expect a large share of modern search traffic to arrive with no query attached at all, because the engines withhold it; that is a limit of web analytics rather than of this node. Rows come back ordered by value descending and cut off at limit, so a result is a TOP N rather than the whole picture: raise limit or page through before treating it as complete, and never add two periods' top lists together, because the top 20 of January plus the top 20 of February is not the top 20 of both. Output shape: data.reports is Clicky's own response verbatim, one entry per requested type, each holding dates and then items. data.rows is a flat view DERIVED by this connector, where every item carries its own type and date as real columns, because Clicky puts both on the row's parents rather than in the row. Bind charts and datasets to data.rows. Plan caps, from Clicky's own docs: a FREE account can only reach the last 30 days of history and returns at most 100 results per request, so a longer window or a bigger limit quietly returns less than you asked for. The date range itself is capped at 366 days on Pro and 31 days on every other plan.
Clicky: Top Pages
Your most-viewed content for a period, one row per URL with its page views, its share of the total, its title and a link into the Clicky dashboard. Also serves site-domains, which splits the same traffic by which domain name it arrived on when several domains report into one site ID. Use Item to pull a single page's number, or Filter to narrow to a section such as /blog. Rows come back ordered by value descending and cut off at limit, so a result is a TOP N rather than the whole picture: raise limit or page through before treating it as complete, and never add two periods' top lists together, because the top 20 of January plus the top 20 of February is not the top 20 of both. Output shape: data.reports is Clicky's own response verbatim, one entry per requested type, each holding dates and then items. data.rows is a flat view DERIVED by this connector, where every item carries its own type and date as real columns, because Clicky puts both on the row's parents rather than in the row. Bind charts and datasets to data.rows. Plan caps, from Clicky's own docs: a FREE account can only reach the last 30 days of history and returns at most 100 results per request, so a longer window or a bigger limit quietly returns less than you asked for. The date range itself is capped at 366 days on Pro and 31 days on every other plan.
Clicky: Uptime Checks
Clicky's uptime monitoring, if it is enabled on the site: uptime is the overview of checks and uptime-list is the event log of individual outages and recoveries. Uptime monitoring is a paid Clicky feature, so an account without it gets an empty report rather than an error, which is worth handling in a workflow that alerts on the result. Output shape: data.reports is Clicky's own response verbatim, one entry per requested type, each holding dates and then items. data.rows is a flat view DERIVED by this connector, where every item carries its own type and date as real columns, because Clicky puts both on the row's parents rather than in the row. Bind charts and datasets to data.rows. Plan caps, from Clicky's own docs: a FREE account can only reach the last 30 days of history and returns at most 100 results per request, so a longer window or a bigger limit quietly returns less than you asked for. The date range itself is capped at 366 days on Pro and 31 days on every other plan.
Clicky: Visitor Locations and Organizations
Who your visitors are and where they connect from: countries, cities and regions, the languages their browsers ask for, and the network side of the same question, hostnames and organizations (which is how you spot a specific company reading your site). Organization and hostname data is derived from the visitor's IP and is missing or wrong for a large share of consumer and mobile traffic, so treat it as a lead rather than a fact. Rows come back ordered by value descending and cut off at limit, so a result is a TOP N rather than the whole picture: raise limit or page through before treating it as complete, and never add two periods' top lists together, because the top 20 of January plus the top 20 of February is not the top 20 of both. Output shape: data.reports is Clicky's own response verbatim, one entry per requested type, each holding dates and then items. data.rows is a flat view DERIVED by this connector, where every item carries its own type and date as real columns, because Clicky puts both on the row's parents rather than in the row. Bind charts and datasets to data.rows. Plan caps, from Clicky's own docs: a FREE account can only reach the last 30 days of history and returns at most 100 results per request, so a longer window or a bigger limit quietly returns less than you asked for. The date range itself is capped at 366 days on Pro and 31 days on every other plan.
Clicky: Visitor Segmentation
Ask one report about one slice of your audience: top searches for visitors from Germany, top pages for people who arrived on a campaign, entry pages for returning visitors only. Segments names WHICH reports you want back and the filter fields below define WHO is in the segment; both halves are required for a useful answer, because an empty segment is just the ordinary report. Clicky caps segmentation at a 31-day range and 1000 items on every plan, which is tighter than the ordinary reports and is not a free-tier limit you can pay past. Output shape: data.reports is Clicky's own response verbatim, one entry per requested type, each holding dates and then items. data.rows is a flat view DERIVED by this connector, where every item carries its own type and date as real columns, because Clicky puts both on the row's parents rather than in the row. Bind charts and datasets to data.rows. Plan caps, from Clicky's own docs: a FREE account can only reach the last 30 days of history and returns at most 100 results per request, so a longer window or a bigger limit quietly returns less than you asked for. The date range itself is capped at 366 days on Pro and 31 days on every other plan.
Clicky: Visitor Totals
How many people visited: total visitors, unique visitors, first-time visitors, and how many are on the site right now. Turn Daily on to get one row per day for a traffic chart. Aggregation warning, and it is the one that produces wrong dashboards: NONE of these numbers may be added across days. Unique and new visitors are deduplicated over whatever range you asked for, so thirty daily uniques is not the monthly unique count, and visitors-online is a live gauge that ignores the date range entirely. Ask Clicky for the longer period instead of summing shorter ones. Output shape: data.reports is Clicky's own response verbatim, one entry per requested type, each holding dates and then items. data.rows is a flat view DERIVED by this connector, where every item carries its own type and date as real columns, because Clicky puts both on the row's parents rather than in the row. Bind charts and datasets to data.rows. Plan caps, from Clicky's own docs: a FREE account can only reach the last 30 days of history and returns at most 100 results per request, so a longer window or a bigger limit quietly returns less than you asked for. The date range itself is capped at 366 days on Pro and 31 days on every other plan.
Clicky: Visitor and Action Log
Session-level detail rather than totals: visitors-list is one row per visit with time on site, referrer, browser, operating system, geolocation and any custom data; actions-list is one row per page view, download, outbound click or event; visitors-most-active ranks the people who came back most often. This is the node behind a real-time alert workflow, usually with Time Offset set to the last few minutes rather than a date. It is also the node that returns PERSONAL DATA (IP addresses, hostnames, session identifiers), so route its output deliberately. Clicky caps these lists at a 31-day range and 1000 items on every plan, and at 100 on a free account. Output shape: data.reports is Clicky's own response verbatim, one entry per requested type, each holding dates and then items. data.rows is a flat view DERIVED by this connector, where every item carries its own type and date as real columns, because Clicky puts both on the row's parents rather than in the row. Bind charts and datasets to data.rows.
Clicky: Web Browsers
Which browsers your visitors used, ranked by visits, with each browser's share of the total. Use Item with an exact browser name to track one browser over time, which is the usual reason to automate this report rather than read it in the dashboard. Rows come back ordered by value descending and cut off at limit, so a result is a TOP N rather than the whole picture: raise limit or page through before treating it as complete, and never add two periods' top lists together, because the top 20 of January plus the top 20 of February is not the top 20 of both. Output shape: data.reports is Clicky's own response verbatim, one entry per requested type, each holding dates and then items. data.rows is a flat view DERIVED by this connector, where every item carries its own type and date as real columns, because Clicky puts both on the row's parents rather than in the row. Bind charts and datasets to data.rows. Plan caps, from Clicky's own docs: a FREE account can only reach the last 30 days of history and returns at most 100 results per request, so a longer window or a bigger limit quietly returns less than you asked for. The date range itself is capped at 366 days on Pro and 31 days on every other plan.
Connecting Clicky takes one credential
Both values come from Clicky's site preferences page for the site you want to report on, which is where its API documentation says the Site ID is available. Take the ordinary sitekey, which Clicky describes as a 12 to 16 character string unique to each website. Clicky also keeps an admin sitekey on the main preferences page for a site, and that one is the wrong key here: Clicky's manual tracking documentation describes it as the key that allows you to write data to Clicky and says not to share it with anyone else. This connector only reads, so it never needs the admin key and there is nowhere to put it. If you have more than one site in Clicky, each has its own pair and the two halves are not interchangeable, so make sure the Site ID and sitekey you copy belong to the same site. That is the other half of the most common first failure, and it presents identically to a wrong key.
Sign in to Clicky and open the site preferences page for the site you want to report on. Copy the numeric Site ID and the sitekey, and leave the admin sitekey alone.
Decide which site this credential is for. A Clicky sitekey belongs to exactly one site, so a second site needs a second stored credential rather than a second node setting.
In Mattering Automations, open Connectors, click Add Connector and choose Clicky, then Create new credential and paste the Site ID and the sitekey.
Click Test Connection. Mattering runs the cheapest authenticated read on the API, a single visitor tally for today, which proves the two halves are a matching pair and that the account can read stats. A rejection here almost always means the admin key was pasted, or the Site ID and sitekey came from different sites.
Read what the success message adds, because it is the part people skip: a passing test does not reveal which Clicky plan the account is on. A free credential passes and then returns 30 days of history and 100 rows per request from then on, silently, because Clicky truncates rather than erroring.
All 18 Clicky nodes appear in the canvas palette. Drop Clicky: Visitor Totals on the canvas and run it as it comes, then turn Daily on before you chart anything, because without it a multi-day request returns one undated aggregate instead of a row per day.
Common uses for the Clicky MCP
A weekly traffic digest that does its arithmetic honestly
Clicky: Visitor Totals with Daily turned on gives one row per day, Clicky: Action Totals gives the page views and downloads beside it, and Clicky: Top Pages gives the content behind the number. An AI step can then write the paragraph a person actually reads. The trap this connector marks for you is in the maths: action counts are event counts and add up cleanly across days, but unique and new visitors deduplicate over whatever range you asked for, so seven daily uniques summed is not the week's unique visitors on any analytics platform. Ask Clicky for the week rather than adding its days together, and the node descriptions say which is which so an agent building the workflow can read it too.
Real-time alerting on the visitor log
Clicky: Visitor and Action Log has a Time Offset field that returns only the last N seconds of activity, up to a day, which turns it into a poller: set 300, run it every five minutes, and act on what appeared. That is how you get a message when someone from a target account lands on the pricing page, or when a campaign starts converting. Two cautions belong in the design. This node returns personal data, including IP addresses and session identifiers, so route its output deliberately rather than dumping it into a public channel. And Clicky allows one simultaneous request per IP address per site ID, so a poller that fans out across several Clicky nodes at once will collide with itself; run them in sequence.
Watching for new coverage and new referrers
Clicky: Referrers and Traffic Sources can request the links-unique report, which lists referrers seen for the very first time rather than the ones ranked highest, and that is the difference between a backlink watch and a leaderboard you already know by heart. A scheduled run that compares each day's first-time referrers against what you have already recorded turns a press mention or a forum thread into a notification on the day it happens. Pair it with Clicky: Top Pages filtered to the section that got linked, so the alert carries what the visitors actually read when they arrived.
An SEO report an agent can assemble on demand
Clicky: Searches and Keywords covers the whole search side in one node: full queries, the queries split into keywords, which engines sent the traffic, local search terms, and Clicky's own view of your Google rankings ordered best to worst. Combined with Clicky: Top Pages and Clicky: Entry and Exit Pages, an assistant can answer which posts are earning search traffic and which landing pages are losing people, without anybody opening a dashboard. Set expectations on one point: a large share of modern search traffic arrives with no query attached at all because the engines withhold it, which is a limit of web analytics rather than of this connector.
Loading daily analytics into a database or a dataset
Every node returns data.rows, a flat view this connector derives where each row carries its own type and date as real columns, which is exactly the shape a Mattering database or an Analytics dataset wants. Turn Daily on so each row gets a real date, run the node on a schedule, and you have a history that outlives Clicky's own retention window on your plan. Two details make the difference between a table and a mess: the ranked reports return a top N rather than everything, so store what you fetched rather than treating it as the full picture, and the Null Values option on Clicky: Run Any Report Type keeps empty fields in the response so your columns stay stable instead of going ragged.
Answering a segment question without building a report
Clicky: Visitor Segmentation is the node for questions with a WHO attached: top pages for visitors from Germany, entry pages for returning visitors only, searches for people who arrived on a specific campaign. You name which reports you want back and which filters define the segment, and Clicky computes the intersection. This is the node an assistant reaches for when a question arrives in chat rather than in a spec. Clicky documents a tighter set of restrictions on segmentation than on the ordinary reports, a maximum date range of 31 days and a maximum of 1000 items per request, so keep the question inside a month and page for the rest.
Example: A Monday morning traffic report that lands before the standup
Built on the Automations canvas in a few minutes, or described in plain language to Matter and built for you.
Trigger: a weekly schedule fires. This connector has no trigger nodes, because Clicky's API documents no webhooks or HTTP callbacks, so Clicky workflows are scheduled or invoked rather than pushed.
Clicky: Visitor Totals runs with Daily turned on for the previous seven days, returning a row per day for the chart rather than one undated aggregate.
Clicky: Action Totals runs over the same window for page views and downloads, which are event counts and may safely be rolled into a weekly total.
Clicky: Top Pages and Clicky: Referrers and Traffic Sources run for the same window, so the report carries what people read and where they came from instead of only a line going up or down.
An AI step reads data.rows from each and writes the summary, with the daily unique visitor figures presented as a series rather than summed, because uniques deduplicate over the range they were asked for.
The result is posted to a channel and filed as a Mattering Productivity task, so the week's numbers are somewhere a person can comment on them rather than only in a chat scrollback.
Clicky MCP questions, answered
What can this connector actually do in Clicky?
18 nodes, all of them reads. Visitor totals including who is on the site right now, action totals covering page views, downloads, outbound clicks and events, engagement and time figures including bounce rate, top pages and site domains, entry and exit pages, referrers and traffic sources including first-time referrers, the whole search side including keywords, engines and rankings, visitor locations and organizations, browsers, operating systems and devices, screen resolutions, outbound links, downloads, events and video, goals, campaigns, split tests and custom data, visitor segmentation, the session-level visitor and action log, uptime checks, one alert by its id, and a catch-all node that accepts any of the 57 documented report types. What is not here, because Clicky's API does not document it or does not offer it to this credential: writing anything at all, creating or deleting a site, creating or editing a goal, deleting recorded data, listing your sites, listing your goals, listing your alerts, funnel, cohort or retention analysis, and webhooks or HTTP callbacks of any kind.
Does this work with Claude, ChatGPT, or other AI assistants?
Yes. Mattering exposes these nodes over MCP, an open protocol for connecting AI assistants to tools, so any MCP-capable assistant, including Claude and ChatGPT, can connect to your workspace and read Clicky through them. Matter, the assistant built into the platform, uses the same nodes natively. The same nodes are also steps on the automation canvas, so a report an assistant helps you assemble in a conversation can then run every Monday with nobody present. Because every node in this connector reads and none writes, handing an assistant the whole Clicky surface carries less risk than most connectors; the thing to be deliberate about is not damage but disclosure, since Clicky: Visitor and Action Log returns IP addresses and session identifiers.
Does Clicky publish its own MCP server?
No. Clicky's help documentation covers the stats API, the account API and the white label API and does not mention MCP or the Model Context Protocol anywhere, and no Clicky MCP package is published under a Clicky or Roxr Software name. There is a community one, clicky-mcp by Colin Toh on GitHub, which is an independent project rather than a vendor product; it holds the same site id and sitekey in a local environment file and does the job it sets out to do. The difference worth weighing is not the chat window, where a local server is a perfectly good answer, but what happens when the chat window is closed. These nodes are also steps on an automation canvas, so they run on a schedule or from another workflow's output with nobody present, chain with every other connector in one workflow, hold the credential once in a shared workspace rather than in each person's config, and carry the aggregation rules per report family so a scheduled job does not quietly sum figures that must not be summed.
Which key do I paste, and where is it?
Two values, both from the site preferences page in Clicky for the site you want to report on: the numeric Site ID, which Clicky's API documentation says is available there, and the sitekey, which Clicky describes as a 12 to 16 character string unique to each website. Take the ordinary sitekey, not the admin sitekey. Clicky keeps an admin sitekey on the main preferences page for the same site, describes it in its manual tracking documentation as the key that allows you to write data to Clicky, and says not to share it with anyone else. That is why pasting it here is the most common first failure, and also why this connector never asks for it: nothing in it writes. The other common failure looks identical at the point it fails: a Site ID and a sitekey copied from two different sites. A Clicky sitekey belongs to exactly one site and is meaningless against another site's id, so both halves have to come from the same place. Test Connection catches both.
Can I use this on a free Clicky account?
Yes, and it is worth being exact about why. API access is not one of Clicky's premium features: its pricing page lists API access among the basics that every plan gets, including the free one. So you can connect and run every node. What the free plan changes is what comes back. Clicky's API documentation states that free accounts are limited to the last 30 days of history and 100 results per request, and its pricing page states that free plans keep 30 days of history while paid plans keep it indefinitely, with individual visitors and actions guaranteed for six months. Clicky enforces those by truncating rather than by returning an error, which is the part that costs people an afternoon: a request for a quarter looks healthy and returns a month, and a top 500 request returns a top 100. Two report families are gated by feature rather than by cap. Goals and revenue are premium features the free plan does not include, and uptime monitoring is listed as Pro Plus and above, so base Pro does not have it either. Without them there is nothing for Clicky: Goals and Campaigns or Clicky: Uptime Checks to return, so a workflow that alerts on the result should handle an empty report as its own case rather than as a zero. None of this shows up in Test Connection, because a free credential is a perfectly valid credential.
What are Clicky's rate limits?
Clicky's API documentation states one limit plainly: a maximum of one simultaneous request per IP address per site ID, which it says can be removed with a custom plan. That is a concurrency limit rather than a requests-per-minute quota, and it has a specific consequence for automation design: a workflow that fans several Clicky nodes out in parallel collides with itself, so run them in sequence. The other numbers shape what one request may ask for rather than how often you may ask. The row limit is 1 to 1000, or the literal word all for up to 5000, with a vendor default of 50. The date range is capped at 366 days for Pro users and 31 days for everyone else. And Clicky documents a separate set of restrictions on the visitors-list, actions-list and segmentation reports specifically: a maximum date range of 31 days and a maximum of 1000 items per request, with the page parameter used to fetch further batches of 1000.
Why are there typed report nodes instead of one node with a type box?
Because a free-text parameter box would have made every mistake silent. Clicky's whole reporting surface is one endpoint parameterised by a type value, and it has 57 of them; a node that accepts any string returns an empty report for a typo, a wrong guess and a genuinely empty week alike. Each node here instead owns a report family, offers only the values that belong to it, and validates them before the call, so a wrong value fails with a message naming the ones it accepts. The grouping is also what makes the aggregation rules truthful: every type Clicky: Action Totals can request is an event count, which is why that node can declare its numbers additive, and the averages and rates that look like they belong beside them live on Clicky: Engagement and Time Totals instead, where nothing is declared summable. The mapping is a partition, so all 57 values are reachable, and Clicky: Run Any Report Type takes any of them when our grouping is not yours.
Which Clicky numbers can a workflow safely add up?
Actions can, visitors cannot, and rates never can. Everything Clicky: Action Totals returns is an event count, so a fortnight really is the sum of its fourteen days and a dashboard may roll days into weeks and months freely. Clicky: Visitor Totals is the opposite: unique and new visitors are deduplicated over whatever range you requested, so thirty daily uniques is not the month's unique visitors, and the live count of people on the site right now ignores the date range entirely. Ask Clicky for the longer period instead of summing shorter ones. Clicky: Engagement and Time Totals declares nothing summable at all, because bounce rate and average time are rates, an unweighted mean of daily rates is also wrong, and Clicky returns no visitor count beside them to weight by. And every ranked report is a top N: the row limit decides which rows exist, not just how many are shown, so January's top 20 plus February's top 20 is not the top 20 of both months.
Put Clicky to work inside your workspace
Connect once, then use every node in visual automations and AI agent workflows across the Mattering platform.
All product names, logos, and brands are property of their respective owners and are used for identification purposes only. Use of these names, logos, and brands does not imply endorsement or affiliation. Clicky is a product of its respective owner; Mattering is an independent platform.