Features

Add slash commands to control Robo-Rewsty’s behaviour and @ mentions to target specific Rewst objects. For example: Slash Commands /mode — Switch modes: architect, builder, debugger, reviewer, optimizer, teacher. /grill-me — Question me until requirements, edge cases, failures, dependencies and unknowns are fully understood. /challenge — Challenge the design and assumptions. /red-team — Find every possible production failure. /blueprint — Turn requirements into workflow architecture. /plan — Create the implementation plan. /build — Start building. /audit — Full workflow quality/security review. /debug — Trace symptom → cause → fix. /trace — Follow a variable/value through the workflow. /explain — Explain a workflow or component. /simplify — Reduce unnecessary complexity. /optimize — Improve performance/API efficiency. /refactor — Improve structure without changing behaviour. /jinja — Jinja specialist mode. /api — API integration specialist mode. /edge-cases — Find edge cases. /errors — Review failure/error handling. /test — Generate test scenarios. /simulate — Simulate a workflow execution. /what-if — Test hypothetical scenarios. /document — Generate technical/support/handoff docs. /summarize — Summarise decisions, architecture and next steps. /checkpoint — Capture the current design state. /next — Recommend the best next action. /todo — Show remaining work. /production-check — Final go-live review with Deploy: YES/NO + why. @ Mentions Reference Rewst objects directly: @GetUser · @action:GetUser · @workflow:UserOnboarding · @subflow:CreateUser · @trigger:Webhook · @transition:UserExists · @var:user_email · @alias:customer · @orgvar:Variable · @integration:MicrosoftGraph · @form:Request · @section:UserCreation Examples: /debug @GetUser /trace @var:user_email /audit @workflow:UserOnboarding /red-team @workflow:Offboarding /what-if @MicrosoftGraph returns 429 Create Objects Support: @new workflow · @new subworkflow · @new action Example: Refactor @GetContact @ValidateContact @MatchCompany into @new subworkflow. Command Chaining Allow commands to work together: /grill-me /blueprint /audit /simplify @workflow:UserOnboarding /audit /red-team /production-check @workflow:Offboarding

AI-powered crates such as Rewst's AI PSA Analyzer let you choose the AI provider on the setup form: Rewst's internal AI, or your own OpenAI or Anthropic integration. Rewst already ships a Hatz AI integration that exposes OpenAI-compatible chat completions and a list-available-models action, so partners who have standardized on Hatz AI internally already have the credentials and the connection in place. They cannot point an AI crate at it, which means either running AI spend through a second provider they do not otherwise use, or cloning the crate to swap the provider call themselves. Adding Hatz AI to the provider selector on AI crates would let those partners keep AI usage consolidated on the platform they already pay for and govern. Starting with the AI PSA Analyzer would cover the most common request, but the same selector appears on the other AI crates too.

Today the only way to test a GraphQL query against the Rewst API is to build or open a test workflow, add a Generic GraphQL request action, execute the workflow, and read the result back out of the execution. That is several steps for what is often a one-line change while iterating on a query. GraphQL has no table-style schema to browse, it is class based, so schema discovery is itself trial and error: you frequently need to run a query just to find out what fields or accepted argument values exist. A live GraphQL tool, in the spirit of the existing Live Editor, where you type a query, hit execute, and get the response immediately, would let builders confirm a query is structured correctly before it ever goes into a production workflow. The browser developer console is a partial workaround but adds its own steps. Anyone doing advanced platform work (querying forms and labels, building App Builder pods against platform data) hits this repeatedly.

Add azure devops as source control option similar to github source control Documentation for the API can be found here: https://learn.microsoft.com/en-us/rest/api/azure/devops/?view=azure-devops-rest-7.2 I would love to be able to use this in the same way as github

It would be very helpful if a workflow had a button that showed a list of other workflows that call this workflow. There are occasions where a call-breaking change might be made to a small component who is reused frequently and currently it's necessary to wait for failures and process through those to remediate in the absence of manual documentation.

Sometimes we want to run an action/task in a workflow as the top level org. Currently, it can only run as the running org or anything below the current org

Right now when looping through a list with items, if one of the items causes a failure the whole process stops, fails and exits out. I would want an 'on error continue' option so that the loop finishes and I can review the errored items and decide on the next action.

Create a condition before a trigger, specifically webhooks, which can be enabled or disabled, and will basically insist that the workflow can only do one alert or whatever is gathered from the webhook at a time. With that being said, if 10 alerts came through at once, then the webhook, would go through all 10, one by one. Thanks team

A hosted Python execution environment, similar to the existing PowerShell Interpreter, would allow workflows to run native Python code (e.g. via a #py block) without relying on external infrastructure.

Often, a task fails for some unknown reason (particularly those from integrations) and a simple retry does the trick. However, building a retry loop into every task is quite a hassle and with crate workflows, it means un-syncing or having to use a wrapper workflow and retry the entire crate workflow. Having a built-in task/action retry parameter would help a lot.