Agent Studio

I have a compound action:steps:  - for:      output_key: user_bookings      each: booking      in: data.list_of_user_bookings      index: user_index      steps:        - try_catch:            try:              steps:                - action:                    action_name: mw.get_user_by_email                    output_key: got_user                    input_args:                      user_email: booking.email                - switch:                    cases:                      - condition: data.lang == "es"                        steps:                          - notify:                              output_key: officespace_output1                              recipient_id: data.got_user.user.id                              resources:                                - name: instructions                                  dataty

Hi , I have a use case where I am using compound action to create buttons, i would like to give links to these button.orAre there any way to display the buttons with link using html tags. 

The Purple API Tool (https://docs.moveworks.com/agent-studio/quickstart-guide/purple-api-tool) has been broken since 3/17/2026. What's the timeline of this Purple API Tool being fixed?Now with the same session ID, webhook url and my API Key filled properly in all three fields, and webhook payload generated, when I clicked Trigger Webhook button, I kept getting "Error in API: An error occurred. You may be missing the API key in your input".

Lighter session this week, but we still covered ground. Ryan and I walked through the new image delivery feature in Notify, unpacked when to reach for Python scripts over DSL, and ran through a chunk of the Agent Studio roadmap — Agent Architect V2 and MCP support included.What we covered:Media Delivery (Image URLs) — send images to users via Notify's new image_urls field Python Scripts vs DSL — when to pick Python over mappers for string building, nested data, and math Plugin Export & Dev-to-Prod — export flow walkthrough plus versioning on the roadmap Verbatim Messages & Embedded Links — render templates with markdown formatting for inline links Agent Studio Roadmap — Agent Architect V2 vibe coding, one-click connectors, sharing & permissionsHere's the recording:

HI ,Any plans with an option to create the folders is Agent Studio platform so that we can manage our use cases in folder structure. Regards,Rajeev

 Hi , I am sending the teams nudge using moveworks send_message API, the nudge only works for 1 email id not for multiple. can someone please help here. following is the code for the reference.        var recipientEmail = event.parm1;// value being passed in event.parm1 is -> email1@example.com,email2@example.com        var summary = event.parm2;// nudge works with following hard coded values    //    emailRecipient = ["email1@example.com", "email2@example.com"];// following code is to get the list of string        var colSplit = recipientEmail.split(",");// logging result is [colSplit 1 -- object] gs.log("colSplit 1" + "--" + typeof colSplit); var arr = []; for (i = 0; i < colSplit.length; i++) { arr.push(colSplit[i]); }       try {

Hey Everyone,So I know when it comes to the AI Assistant, the graph.api access needs to be on the appliation level.That said, when it comes to agent studio can it work with graph.api access on the delegate level? We’re running into some issue with the access through our Global tenant and if we can use delegate access with Agent Studio that would open up a lot more doors for us faster.

I am using a listener to take in requestIds that start an automated process. When setting it up, there were some situations where the requestId was sent twice and I wanted to avoid the processing being called twice, so I added in Deduplication configuration in the listener. The request body being sent to Moveworks is always in this format:{“requestId” : “<id>”}Therefore for the deduplication, I added requestId as the Deduplication key (no spaces or quotes) with a 1 min window. However I am noticing that it is blocking requestIds that are not equal but are sent within the 1 minute timeframe. The difference between them in these cases is the last character in the requestId (ex ABC1234 vs ABC1235) In the logs, I am seeing one go through and “debounced_update” log entries for the other requestIds sent. I have no event filtering enabled for the 1 plugin that is associated to this listener. Please assist me in understanding the root issue and how to set this up correctl

Is it possible to display a QR code to a user via the bot? The agent architect makes it seem like there is but it’s suggestions do not work. If not, are there any roadmap items to include image displays via the bot?

In my application, all request bodies need to be form data with data as the key and the json body as the value. I am able to handle this in http actions with raw json body by having it be something like data={<insert body here>} and having the header of Content-Type=application/x-www-form-urlencodedI want to use the new data mapper functionality in http actions so it will handle escaping and such for me, however I cannot seem to find a way to get it to work with this setup. I have attempted numerous times with variations of data: <json to mapper syntax> , but the app always returns with a missing data element error. Is there any way to handle this currently?

Hey builders 👋I've been reviewing a lot of plugins lately, and the same slot mistakes keep showing up. They all pass in testing. They all break in production. So I wrote up the six patterns I see most often, with the bad config and the fix for each. 🚨 1. Using string for peopleThe #1 slot mistake I see. You're asking the engine to parse emails from free text.# Bad — forces the engine to parse structured data from free text- name: attendees description: "Comma-separated list of attendee emails like john@company.com, jane@company.com" data_type: stringUser says "invite John and the marketing team." Your string slot captures the literal string "John and the marketing team". The engine tries to parse that into emails. Fails. User tries "john@company.com, marketing-team@company.com". Engine parses it. But marketing-team is a distribution list, not a user.

This may not be related to Agent Studio specifically but it relates to a plugin I’m creating that prompts for user OAuth2.0 authorization so I thought I’d post here first.I'm looking for a way for users to be able to clear Moveworks authorization tokens after they've already made an authorization between their Moveworks account and some other application. For example, if a user has made an authorization between their Moveworks account and Jira, the user can remove their Moveworks connection in their Jira settings, but I don’t see a way for users to remove the Jira connection in Moveworks. With Moveworks still seeing the connection which is no longer valid, the plugin then returns authorization errors and doesn’t prompt for a new authorization to overwrite the existing authorization.I didn't find any documentation on this, and the instructions I'm given by Agent Architect do not work, for example asking the bot to manage connections or disconnect the application. Agent

I’ve built a plugin that sends notifications when Jira approvals are needed. The approvers of the Jira ticket receive the notification and have buttons to approve or deny the Jira ticket. In order to approve or deny, the user must first authenticate a Moveworks Connector with their Jira (Atlassian) account. This all works fine, but after authenticating, the approve and deny buttons no longer work because the information that was stored in them is wiped. Is there a way to work around this so the information in the buttons are not wiped? Or to send another set of buttons to the user after authenticating? Or some other workflow to accomplish this?Here is how the plugin is built..Jira ticket sends information to a Moveworks webhook listener when an approval is required. Plugin is triggered with the webhook and runs a compound action that sends a notification to the approvers about the Jira approval, and also included  are approve and deny buttons to action on

I have the following built-in connectors, and I have no customized connectors yet.ms_graph    Microsoft Graph (built-in connector)msgraph_teams    Microsoft Graph (built-in connector)msteams    Microsoft Teams (built-in connector)And I’m working setting up the following plugin https://marketplace.moveworks.com/plugins/outlook-lookup-availability. However I noticed that I already have the Microsoft Graph built-in connector, therefore I was wondering if I need to set up a customized connector even thought I have an existing pre-built one. Or if I need to configure a customized connector to build the mentioned plugin? The reason why I’m asking this, it’s because am getting different answers when using Moveworks Agent Architect

Hello Community!I wanted to share a set of practical use cases for managing device health that empowers end-users to take action when their hardware needs attention. While these notifications and workflows leverage Nexthink, other device management tools can be substituted using a similar architecture.The Solution: Nexthink Ambient AgentsNexthink offers a Remote Actions API to initiate tasks, native webhooks for real-time alerts, and Nexthink Query Language (NQL) to define data triggers. Combining these capabilities allows ambient agents to listen, notify, and assist users with actions such as:Clear Device/Teams Cache Restart Device

We redesigned the Developer Hub into the single page I wish existed when I was learning Agent Studio. Quick links to basically everything you need: docs, quickstarts, cookbooks, marketplace, API reference, and more.Everything you already use is still there (Quickstart Guides, Cookbooks, Purple Chat Builder, Academy, extensibility docs). Just easier to find.The goal was simple: one place, no hunting. Building a plugin, troubleshooting a slot resolver, figuring out where to start. Developer hub gets you there in one click.Go poke around and bookmark it: https://www.moveworks.com/us/en/developers you might see a familiar face

Hello Community!While Moveworks supports an Approvals skill out-of-the-box, I wanted to share a common architecture to create custom approvals via Agent Studio! Using Webhooks and Agent Studio, you can create a seamless, "in-channel" approval experience with full customizability that keeps workflows moving without forcing users to log into external portals.The Solution: Webhook-Driven ApprovalsBy leveraging Agent Studio Webhooks, you can process external events to provide users with real-time notifications and actionable buttons to approve or deny requests. Common workflows requiring approvals include:Software Access Requests Hardware Procurement Requests Travel & Expense

Encountering the following issues with Agent Studio Quickstart Guide Exercises: 1) Quickstart #3: Basic Process Agent: Not able to generate/view any Purchase Requests in Moveworks Purple API as shown below in the screenshot. Tried creating multiple session IDs, but this step won’t work. Link to exercise: https://docs.moveworks.com/agent-studio/quickstart-guide/quickstart-basic-process-agent  2) Quickstart #4: Structured Data Analysis (SDA): Facing issues with the “Fetch Accounts” endpoint and receiving the following 401 error:Link to exerc

Has anyone successfully setup the Docusign Connector and installable plugins from the Marketplace? We loaded the installable plugin and believe we have correctly setup the connector using Oauth2 and JWT Auth.The trouble is hitting when we try and run the HTTP Action: Docusign_Get_Account_Id_By_User_Email. I keep getting a 401 error. Confirmed with Docusign support that the consent is setup correctly, but unsure what’s missing. Thought it might be the authorization in the header not being passed, but again unsure as this is our first go at this Plugin. 

Hi Team,We are developing a use case to change the profile picture. This is a pretty straightforward use case where we have a File data type slot and an API call to upload the photo to Workday.Workday does have a couple of validations for an uploaded photo that we are trying to check via Moveworks:1. Check whether the uploaded picture is an image (jpeg, png) - we are doing this by checking the file name in the validation of the slot.2. Check the file size. (probably less than 1MB)3. Check if the image is a square or a certain aspect ratio.4. Can we validate the optical recognition - if the uploaded picture is a passport, a driver's license, etc.?I asked the Agent Architect bot, but the recommendation it gave wasn't possible (a script action for the file size).

I have a notify step that uses RENDER() with a generated input string training_str. The template uses HTML tags (e.g. <b>) and a training_str built from a $MAP / $CONCAT expression, but the newline characters render literally (you see \n) when the message is sent to Teams. I’ve also tried "<br>" and "<br >" and they also end up displayed literally. How can I generate a string from my inputs that will render new lines appropriately in Teams? MS Teams screenshot of how the message renders with a literal \n Notify YAML (relevant parts only): - notify: output_key: notification recipient_id: data.user_mw_record.user.record_id

Hey builders,Steve dropped the Structured Data Analysis GA announcement, then the session turned into a resolver strategy clinic with invoice lookups, custom object slots, and an LLM-powered ticket routing architecture. Ryan co-hosted, Steve did the heavy lifting.What we covered:Structured Data Analysis (SDA) Now GA — code interpreter for large datasets, auto-triggers at 7K tokens Resolver Input Arguments — fixing dynamic resolvers that return the entire table Custom Object Slots — collecting structured data like attendee email + required/optional flag Output Mappers — stripping raw API noise so the reasoner shows only what users need LLM Actions for Ticket Routing — full webhook-to-assignment architecture with generate structured value New Agentic AI Docs — LLM fundamentals, context windows, and