August 2026

Shopify requires apps published on its App Store to use the GraphQL Admin API, so a store you connect with OAuth now runs the node's GraphQL operations and all 7 of its triggers. That covers the blog and article operations, the prebuilt queries for products, orders, customers, inventory, collections, and fulfillment orders, and Execute Custom Graphql Query and Execute Custom Graphql Mutation, which reach the rest of the Admin API within the permissions you approved.

The node's older operations still run on a store connected with an Access token, or with OAuth through your own Shopify app. Pick one of them on a store connected with NoClick's app and the operation stops and asks you to choose a GraphQL operation or connect your own app.

The six blog operations moved across as part of this, so List Blogs, List Blog Articles, Get Blog Article by Id, Create Blog Article, Update Blog Article, and Delete Blog Article now run on a store connected with OAuth.

Fourteen operations left the Shopify node: the five Price Rule operations, the three Discount Code operations, the five Gift Card operations, and Create Draft Order with Graphql. Each one needed a store permission NoClick does not ask for, so every call returned a permission error instead of doing anything. They no longer appear in the operation picker, and the node now lists 102 operations. A workflow still set to one of them needs a different approach.

Opening NoClick from the admin of a store that installed it from the Shopify App Store now shows a NoClick page inside Shopify. It names the store, shows the connection as Connected and Ready for workflows, and has an Open NoClick workspace button for building and running the workflows themselves.

You can now install NoClick on a Shopify store starting from Shopify. Shopify asks you to approve the store permissions first, then you sign in to NoClick, and the store lands in your account as a connected Shopify credential named after the shop. It is ready to attach to any Shopify node, so you do not have to connect the store a second time from the node's config panel. Installing again on the same store updates that connection rather than creating a second one.

An account that installed NoClick this way stays on the Free plan. Paid plans and recurring credit top-ups are not offered through that channel, so Settings → Billing shows a Free Plan card explaining this instead of the upgrade options, and the upgrade prompts elsewhere in the app say the same. Everything else works as it does on any Free account, including the 25 monthly credits and the 5 workflow limit.

The Shopify node can now work with your store's blog. List Blogs finds the blog to write to, List Blog Articles and Get Blog Article by Id read what is there, and Create Blog Article, Update Blog Article, and Delete Blog Article change it. An article carries a title, an HTML body, an author, tags, a summary, a featured image with its alt text, and whether it is Published or Hidden, so an agent can draft a post and leave it hidden for you to review.

Reading and writing blog content is a new permission, so reconnect your Shopify account before using these operations.

The Store name field now takes your storefront's own domain, such as your-shop.com, or the admin address you copied from your browser. NoClick reads the store handle behind it, so you no longer have to know the my-store part of my-store.myshopify.com before you can connect. If no store can be found for what you entered, the error says where the handle appears in your Shopify admin URL.

The Schedule node's picker now reads as one sentence you can edit in place: "Every 30 minutes · all day · every day · all year". Click all day to set a from and to time, click every day to pick the weekdays it runs on, and click all year to limit it to a range of dates in the month, a range of months, or both. Stacked limits all have to match, so "every 30 minutes, from 9:00 AM to 6:00 PM, on weekdays, from the 1st to the 15th, from March to October" runs exactly that and nothing else.

The words in the sentence adapt to the frequency you picked. Second, minute, and hour schedules take all three limits, a daily schedule takes the day and date limits, and weekly and monthly schedules take the months. Switching frequency drops the limits the new one cannot carry. Every limit clears back in place from the same popover, and the summary line under the picker restates the whole schedule in plain language.

Run times, time windows, and day boundaries are now evaluated directly in the timezone you pick on the node. A 9:00 AM schedule fires at 9:00 AM local time year round instead of moving by an hour when the clocks change.

The popup that opens when a run finishes now walks through the run in reading order instead of listing node outputs one at a time. What came in shows the event that started the run in its own app's frame, so a WhatsApp message looks like a WhatsApp message and an invoice looks like an invoice. What the agent did lists the agent's tool calls with the time each one took and the detail that matters on the row, such as the channel a message went to, and any row opens to show what it was Called with and what it Returned. What went out frames each message the agent sent the way its destination shows it. An agent that worked and chose to send nothing says so in its own words, and a failed run shows the error with a way to fix it instead of a verdict.

The run switcher at the top of the popup reads Latest run and opens your run history: each row has its status, what fired the run, how long it took, how many nodes ran, and the error on runs that failed. It stays reachable while a run's results are still loading and on a workflow that has never run, so history no longer dead-ends.

Type @ in an agent chat and a list of the conversation's workspace files appears. Pick one and its full path is inserted into your message, so you can point the agent at a file without typing the path out. Arrow keys move through the list, Enter or Tab accepts the highlighted file, and Esc closes it.

You can also delete files now. Each row in the Files menu has a trash button that asks you to confirm on the row itself before the file is removed from the workspace.

Every agent on a workflow now has a Linux sandbox with a shell, whether or not you have wired a filesystem node to it. The sandbox starts the first time the agent actually runs a command, so an agent that only ever answers in chat never starts one. In a chat, the agent's /workspace folder is kept per conversation and survives across runs.

A test run answers shell commands with made-up output like every other tool call, so a rehearsal never starts a real sandbox.

The AI builder can look things up for itself while it builds too. Paste an API URL or a sample payload and it can fetch the URL and inspect the real response instead of wiring a node blind. It has no access to your connected accounts while it does this.

You can now upload files into a workspace yourself, so the agent finds them there on its next turn. In an agent chat, use the upload button in the Files menu header, and the files land in that conversation's workspace. In a filesystem node's config panel, use Upload in the file browser, and they land in that node's volume.

Every place you upload a file now shows a percentage and a progress bar instead of a spinner: interface blocks, form file fields, the agent chat composer, the filesystem node's file browser, the chat Files menu, and the media pickers in a node's config panel. An upload that fails says why on its own row.

A single file can be up to 100 MB everywhere, workspace uploads included.

Instagram has a second connection type, Login, which signs you in with an Instagram professional account. Nothing has to be linked to a Facebook Page, and there are no keys to copy. The System user token and Page access token options are unchanged.

Hashtag search, product tagging, and business discovery only exist on the Facebook route, so on a Login connection those six operations stop with a message telling you to connect through Facebook instead.

A Webhook trigger set to When Last Node Finishes used to hand the caller NoClick's own record of the last node's run, with your data buried inside it. Now the last node decides the response. Return an object with a status, an optional headers map, and an optional body, and the caller gets exactly that: a JSON body is sent as JSON, a string body is served as a web page unless your headers say otherwise, and a status with no body works for redirects.

A Serverless Function node that returns a plain string serves it as an HTML page, so a workflow can answer a request with a real REST payload or a real page. Nodes that return anything else keep the previous response.

In the Serverless Function node, an input named data is now just data in your JavaScript, the way it already was in Python. inputs.data still works, and your own const data = ... still takes precedence. Code the AI builder wrote against the documented input names used to stop with ReferenceError: 'data' is not defined.

The home page at noclick.com now shows three parts of NoClick working, in order: an agent handling messages from the apps you already use, the builder turning a prompt into a workflow, and the run record showing every tool call and result afterwards. Each one is the real interface, not a picture of it.

Below them is the list of what NoClick can do and one Start Free button.

Every page under noclick.com/agents now has a Setup section under the hero that walks the whole path before you commit to it: choose whether the agent runs On your computer or In the cloud, pick what it can do in each app, then sign in to each app once.

The last step follows your choice. On your computer ends on the exact command that connects your coding agent to NoClick, shown with a placeholder link because setup mints the real one. In the cloud ends on signing in to the coding agent NoClick runs for you.

• Dropdowns load again. Choosing a Page in Facebook, an ad account in Meta, or one of your posts in Threads failed with "Failed to load options", which left pasting the ID by hand as the only way through.
• Facebook actions that work on a Page, including posts, comments, reactions, insights, Messenger, and webhook subscriptions, now use that Page's own access token. They previously failed with an error about needing a Page access token even when the account had the right permissions.
• Four Facebook actions that could never run were removed: Create Album, Block User, Get Call-To-Actions, and Get Page Tabs. The node keeps 64 actions, including all 11 triggers. A saved workflow still set to one of them will stop with a configuration error.
• When a Meta action fails, the error now carries the reason Meta gives for it, plus the field it objected to and the trace id. An ad that cannot be created because the account has no payment method used to report only "Invalid parameter".
• Localized Items Batch runs again in Meta. The item type it requires is now its own Item Type field rather than something you had to know to add under Additional Fields.
• List Leadgen Forms runs again in Meta. It now fetches the Page's own access token first, which is what that call needs.

The Agent Templates gallery now opens on a search box and a set of filters. Search by job, outcome, app, or node, then narrow by category (Sales, Support, Marketing, Operations, or Engineering), by the apps and nodes a template uses, or by the trigger that starts it.

Every card previews the template's real canvas, drawn from its saved nodes and connections, so you can see the shape of a workflow before opening it. Show more templates loads the catalog in batches, and a search always covers all of it.

Connecting a Zoom account works again. NoClick was sending the list of permissions it wanted to Zoom's consent screen, which Zoom rejects for this app. Permissions now come from the Zoom app itself, so the consent screen loads and the connection completes.

We also removed 416 Zoom actions that could never run: 286 needed Zoom account admin permissions this integration does not hold, and 118 pointed at Zoom endpoints that do not exist. The node keeps 1,224 actions, including all 96 triggers. A saved workflow still set to a removed action now stops with a configuration error naming the problem, where before it failed with an error from Zoom.

Click Test in an agent's chat header to run it against a staged event. Your real agent runs on your real model and prompt, but every tool call it makes is answered with made-up data, so nothing reaches your accounts and nothing is sent. The screen says so up front, and every answered call carries a simulated pill.

Gmail, Slack, WhatsApp, and Telegram triggers ship written situations to test against, and every other trigger gets an editable sample. You can rewrite the staged message in place, keep several named runs per agent, and share a finished run as a read only page anyone can open.

The canvas has a permanent Setup tab (shortcut S) that reads your workflow and lists what it still needs: accounts to connect, variables to fill, actions to pick, which agent runs it, and a Test Run at the end. Two nodes for the same service ask for one account, not two, and connecting once covers both.

Setup ends on something you can act on. For a workflow that hosts an MCP server, that is the connect step for your coding agent with a live check that flips to Connected the moment your client calls in. For everything else, it is a Test Run.

Declare values once under Settings > Variables and reference them from any node config as {{vars.name}}. A bound field shows the resolved value and edits write back to the variable, so you change it in one place.

Mark a variable per user and its value is cleared whenever someone copies your workflow, which turns it into a question their Setup asks. That is how a template can ask a new owner for their repository or their channel instead of quietly running against yours.

Template pages are now built around the agent. Each one previews the real canvas, lists the services setup will ask you to connect and the questions it will ask, and shows the staged conversations the template ships so you can see the agent work before you take a copy. Click Use this agent and your copy opens straight into full screen setup. The gallery is now Agent Templates.

An agent node can hold connected keys for more than one provider, and switching the Model could leave an earlier key attached. That key was used anyway, so the run failed on a rejected or missing API key even when the right key was connected.

NoClick now runs the agent on the credential belonging to the provider of the model you picked. A key for any other provider is ignored, and the run behaves as if nothing were attached: models NoClick serves on its own keys keep running on credits, and models that need your own key stop with a missing credential you can fix in the node's Credentials tab.

When the AI builder asks you to choose between options and more than one can apply, the choices are now checkboxes under a Select all that apply hint instead of letting you pick a single option. Tick as many as you need and the builder receives them all.

Questions where only one answer makes sense still show round options and take one pick, so the form tells you which kind you are answering.

Connecting Cloudflare by signing in failed with a message about an invalid scope. NoClick was asking for one permission Cloudflare does not define, and Cloudflare rejects the whole request at the first permission it does not recognize, so the connection never completed. That permission is gone and the sign-in finishes.

Connecting also asks for offline access now, which is what Cloudflare needs before it will issue the token that keeps a connection working past its first hour. Reconnect an existing Cloudflare credential to pick that up.

These operations returned a routing error on every run and now reach the right Cloudflare endpoint:

• The three Zaraz operations: Get Zaraz Config, Update Zaraz Config, and Publish Zaraz Config.
• All eight Secrets Store operations, covering both the stores and the secrets inside them.
• The four R2 custom domain operations: List, Create, Update, and Delete R2 Custom Domain.
• Get Vectorize Vectors By IDs and Delete Vectorize Vectors By IDs.
• Get Fonts Settings and Update Fonts Settings.
• Get Secondary DNS Config and Update Secondary DNS Config.
• Dismiss Attack Surface Issue.

Three more were reaching Cloudflare but sending a request it would not accept:

• Create Image Direct Upload Url sends a form upload, the only format Cloudflare Images takes for it.
• Create Ai Gateway and Update Ai Gateway send every field Cloudflare requires. Update reads the gateway first and changes only the fields you set, because Cloudflare replaces the whole gateway on each write.

Three operations are gone: Get Account Analytics, Get R2 Bucket Usage Summary, and Get Web Analytics Summary. Cloudflare serves that data only through an analytics API that refuses a credential you create by signing in, so they could not return anything. Cloudflare is now at 591 operations.

A chat bubble now sits in the bottom right of the app and of noclick.com, so you can ask a question without leaving what you are working on. When you are signed in, the conversation carries your email and plan, so we can see which account we are looking at.

• The bubble stays off the pages your own users see: shared agent links, builder input links, public workflow and share pages, credential request links, and embedded canvas previews.
• The Get Started checklist that used to sit in that corner of the canvas is gone, along with the tour that came with it. The walkthroughs for inviting a teammate, publishing, and the agent chat are unchanged.

Asking in Discord still works the same way.

What NoClick asks for when you connect a service is now derived from what each operation actually calls, checked across every service that connects by sign-in. Operations that could only ever fail, because a permission they needed was never requested, now have it. Each integration's page lists the exact set NoClick requests.

A connection keeps only the permissions you granted when you made it, so reconnect the credential to pick up the new ones.

• Jira adds one permission that revives 52 operations across fields, screens, workflows, issue types, and other project configuration.
• GitHub adds the permissions behind its 15 gist operations and its organization reads, 25 operations in total.
• PagerDuty was requesting four permission names PagerDuty does not define. They are replaced with the real ones, alongside event rules and incident workflow instances, covering 22 operations.
• PostHog adds session recording writes, group reads, and organization member reads and writes. Airtable adds record comments, and Sentry adds team administration.
• Slack asks for more permissions too, including the one its On App Mention trigger needs. Tagging the bot in a channel could not start a run before, because the event permission was never requested.

Slack's 104 admin operations now carry an Enterprise Grid badge in the operation picker, and 7 Slack Connect operations an Admin install badge. Slack refuses to install an app that asks for those permissions on an ordinary workspace, so they run on a Bot Token credential from your own Slack app rather than on the sign-in flow. The badge appears when you pick the operation, instead of the operation failing on its first run.

Two operations that a connection you make yourself can never be granted are gone: LinkedIn's Get Post and List User Posts, which need a permission LinkedIn grants to approved partners only, and Dropbox's Permanently Delete File, which is limited to Dropbox Business apps.

145 Cloudflare operations were selectable in the picker but had nothing behind them, so every run returned "Unknown action". They now call the Cloudflare API: addressing and BYOIP prefixes, Intel and Security Center, Workers AI, Log Explorer and Logpull, Magic Transit, Radar AI, Calls and Realtime, Analytics Engine SQL, URL Scanner, Bot Management, and R2 event notifications.

Operations that need a plan you may not have are labeled in the picker: 45 are marked Enterprise, 7 Paid, and 3 Beta.

Triggers that poll on a schedule, like Google Forms On Form Response, Google Sheets On New Row, and Gmail Poll for New Emails, now start watching as soon as their configuration is complete and treat anything that arrives after that as new. Before, the schedule could be created before the trigger had its required fields, which produced a failed run every interval while you were still filling in the panel, and the first poll after that counted everything already there, including the test entry you had just submitted, as history.

• The schedule is created only once the trigger's required fields are filled in, so ticks during setup no longer appear as failed runs in the Logs tab.
• A turn where the AI builder or the MCP server fills in the last required field now registers the trigger, so "your trigger is set up" means it is listening.
• Pointing a trigger at a different form, sheet, or mailbox starts a fresh baseline for the new resource instead of firing for everything already in it. Edits that do not change what the trigger watches, including the ones the AI builder makes while finishing a workflow, no longer reset what it has already seen.
• Running the workflow yourself delivers newly polled items to an AI agent wired to the trigger, the same as a scheduled run. A poll that finds nothing new still runs nothing downstream, manual runs included.
• Changing a node away from its trigger operation removes its schedule, and a trigger whose registration never landed is repaired on its own within a day.

The agent chat in a workflow's Interface tab now lays itself out from the width it actually has, rather than the width of the window. When something else takes up the screen, like the builder drawer or a split view, the chat keeps its full width and Settings opens as a sheet over it instead of splitting the row.

• History, Files, Share, and Settings in the chat header collapse to icons when the header is too narrow for their labels.
• The History, Files, and model pickers open as full-width sheets on phones instead of cramped dropdowns.
• Tablets in portrait use the single-column tab layout, so the navbar and canvas are no longer squeezed into a sliver.

The Honeycomb node brings your observability data into workflows, with 88 operations across events, datasets and columns, boards, queries, markers, triggers, SLOs and burn alerts, and the v2 management API. Connect with a Honeycomb API key or a management key, and pick the US or EU region.

• Two triggers start a workflow when Honeycomb alerts: On Trigger Fired and On Burn Alert. NoClick registers the webhook recipient for you, and can attach it to an existing trigger or burn alert.
• Dropdowns load your real datasets, boards, columns, markers, SLOs, and recipients, so you pick by name instead of pasting IDs.

The Add trigger and Add tool palettes on an agent node now search each service's description, its aliases, and the actions it can perform, not just its name. Searching for "schedule", "timer", or "every day" finds the Schedule trigger, and "create issue" surfaces Linear, Jira, and GitHub.

Services also carry the same name here as on the canvas. Schedule used to be listed as "Trigger Cron" and X as "Twitter", and about a third of the catalog appeared under a name that did not match the one you see everywhere else. Once you pick a service, its list of triggers is searchable by keyword too.

A CLI coding agent's sandbox is billed for as long as it stays up, so a turn that started and never delivered a reply used to keep costing 0.05 credits per minute until the four-hour sandbox limit. Those turns now end: the sandbox shuts down as soon as the agent's process is gone, and a turn that produces no output at all for 30 minutes is treated as stalled and stopped.

• The chat reports the turn as lost and unlocks, so you can send the message again, instead of showing a reply that is still coming and never arrives.
• Files the agent saved to its workspace survive the stop, as they do for any other sandbox shutdown.
• A turn that is genuinely quiet for 30 minutes, for example one long build inside a single command that prints nothing, is stopped as well and has to be sent again.

A photo, voice note, or document sent to a WhatsApp-triggered agent now arrives as a link the agent can download, along with the file type and name. The message caption comes through as the text. Media used to reach the agent as a placeholder with nothing behind it.

• Files up to 25 MB are saved to your workflow's resources at the moment the message arrives, so the link keeps working for the whole run.
• Larger files, and files the provider no longer holds, are reported to the agent as media it could not retrieve, so it can ask the sender to resend instead of guessing.
• Nodes wired downstream of the WhatsApp trigger read the same link from the trigger output.