# Yoda

Agentic Automation. As powerful as the Force. Yoda was originally going to be played by a monkey wearing a mask and carrying a cane, but the idea was scrapped in favor of a puppet created by Jim Henson's Creature Shop and operated by Frank Oz

## NEW

### Meet AI··Task: The new powerhouse of workflow intelligence 💡

Ref: #7170

The AI··Agent is dead 💀

   <div class="docs-screenshot-compact">

![Meet ai task](./assets/meet-ai-task.png)

   </div>

… Long live the smarter **AI··Task**! 🎉

   <div class="docs-screenshot-compact">

![Meet ai task](./assets/meet-ai-task-2.png)

   </div>

:::note
**AI··Task** is a specific, modular task or step that can be embedded within an AI··Agent’s workflow. It represents a concrete _action_ or a _chain of actions_—designed, identified, and optimized by AI for **multi-tool orchestration**. Essentially, while the AI··Agent (e.g. the flow) handles the mission, AI··Tasks are the atomic building blocks that execute each step toward that mission.
:::

---

Follow these steps to configure one AI··Task:

1. **Define the prompt**

   Begin by specifying the **prompt**—this instructs the LLM on the goal or context of the task.

   You can also inject data dynamically into the prompt, enriching the context for the LLM. However, keep in mind that LLMs have token limitations, so overly large inputs may be truncated or impact performance.

   <div class="docs-screenshot-compact">

   ![Define prompt](./assets/define-prompt.png)

   </div>

1. **Add one or more services**

   Unlike the previous AI··Agent (limited to a single service), AI··Tasks can integrate multiple services.

   <div class="docs-screenshot-compact">

   ![Add one more services](./assets/add-one-more-services.png)

   </div>

1. **Configure each service**

   For each service, you can specify:
   - A credential,
   - A custom name
   - A description (to help the LLM understand how to use the service)
   - Authorized operations

   <div class="docs-screenshot-compact">

   ![Authorized operations](./assets/authorized-operations.png)

   </div>

1. **Monitor execution with detailed audit logs**

   Every API call is logged for transparency, traceability, and auditing.

   <div class="docs-screenshot-compact">

   ![Monitor execution detailed audit logs](./assets/monitor-execution-detailed-audit-logs.png)

   </div>

### New Flow design 🎨

We’re excited to unveil a major update in our journey toward a more intuitive and informative design.

   <div class="docs-screenshot-compact">

![Monitor execution detailed audit logs](./assets/monitor-execution-detailed-audit-logs-2.png)

   </div>

Flows are now presented as dynamic cards that deliver richer context at a glance. Each card displays:

- **Title, name, and description** of the flow
- **Deployment environments** where the flow is active (currently limited to one environment—multi-environment support is coming soon 😉)
- **Flow permissions** showing who can interact with the flow

### New Flow permissions 🔐

Based on extensive user feedback and evolving deployment needs, we've undertaken a comprehensive refinement of our permissions model to lay the groundwork for future multi-environment deployments with an enhanced permissions framework

:::note
While multiple-environment deployments are not yet available (coming soon!), we have already refined our framework to accommodate this future feature. For example, the enhanced permission system has been specifically designed to support testing flows in _development_, validating in _staging_, and safely deploying to _production_, with different users having appropriate access levels for each environment and flow. This proactive update to our access control model ensures we'll be ready when multiple-environment deployments launch.
:::

The new permissions system introduces **two distinct sets of roles** to better manage both flow logic and deployment environments. Users can now have different roles for working with flows (Owner, Editor, or Viewer) versus managing environments (Owner, Operator, or Viewer), allowing for more granular access control:

- **Flow Permissions**

  These define who can interact with the flow logic itself.
  - **Owner**: Full control — can edit, delete, and manage flow permissions.
  - **Editor**: Can modify the flow content but cannot delete it or change permissions/credentials.
  - **Viewer**: Read-only access to the flow and associated audit logs.

- **Environment Permissions:**

  These determine access to flow deployments, credentials, and environment configurations.
  - **Owner**: Full access — can manage the environment, its credentials, and deployments.
  - **Operator**: Can deploy flows and view logs, but cannot manage permissions or create credentials.
  - **Viewer**: View-only access to flows, credentials, and deployments within the environment.

   <div class="docs-screenshot-compact">

  ![Environment permissions](./assets/environment-permissions.png)

   </div>

Environment permissions

   <div class="docs-screenshot-compact">

![Flow permissions](./assets/flow-permissions.png)

   </div>

Flow permissions

## FIXES

- Updated documentation for the Export Logs API available here: [Exporting logs](https://docs.mindflow.io/guides/administrating-the-platform/exporting-logs). (#8182)
- MSG to JSON function: missing headers in the final object. (#8176)
- Studio has an invalid cache configuration for HTML documents. (#8240)
- Credentials are lost when I copy and paste a step. (#8109)
- Remove a selection to delete a Note step. (#8078)
- Pasting a step as a new branch following a condition step does not create a new branch but adds the step to the initial branch. (#8059)
- It is not possible to change the Size parameter within the SHA2 function. (#7960)
- Delete playbook dialog doesn’t have the focus. (#7396)
- Change logs are now permanently retained as flow edit history. (#8222)

## INTEGRATIONS

- [AWS Security Incident Response (version: 2018)](https://aws.amazon.com/security/)
- [BeyondTrust PRA Configuration (version: 1.9)](https://www.beyondtrust.com/products/privileged-remote-access)
- [BigPanda (version: 1.0.0)](https://www.bigpanda.io/)
- [Domain Security Scanner (version: 3.0.17)](https://github.com/globalcyberalliance/domain-security-scanner)
- [Dynamics 365 Business Central (version: 1.0.0)](https://dynamics.microsoft.com/en-us/business-central/overview/)
- [Eurecia (version: 1.0.0)](https://www.eurecia.com/en/)
- [Fly Machine (version: 1.0)](https://fly.io/docs/reference/machines/)
- [Intelligence X (version: 1.0.0)](https://intelx.io)
- [Jupiter Ultra (version: 1.0.0)](https://jupiterone.com)
- [Jupiter Trigger (version: 1.0.0)](https://jupiterone.com)
- [Jupiter Token (version: 1.0.0)](https://www.jupiterone.com/)
- [Jupiter Swap (version: 1.0.0)](https://jup.ag)
- [Jupiter Recurring (version: 1.0.0)](https://jupiter.money)
- [Jupiter Price (version: 2.0.0)](https://jupiterapi.com)
- [NetWitness (version: 12.4)](https://www.netwitness.com)
- [Nucleus Security (version: 1.0.0)](https://nucleussec.com/)
- [Office 365 Management Activity (version: 1.0.0)](https://learn.microsoft.com/en-us/office/office-365-management-api/)
- [OTRS (version: 1.0.0)](https://otrs.com/)
- [ProxyPot (version: 3.2.5)](https://proxypot.docs.globalcyberalliance.org/docs/getting-started/)
- [Redis Cloud (version: 1.0.0)](https://redis.com/redis-enterprise-cloud/overview/)
- [SolarWinds Dameware (version: 1.3.01)](https://www.solarwinds.com/dameware)
- [SolarWinds Observability (version: 1.0.5)](https://www.solarwinds.com/observability)
- [SolarWinds Service Desk (version: 0.2.18)](https://www.solarwinds.com/service-desk)
- [TrustBuilder (version: 11-05f0473)](https://www.trustbuilder.com/)
