Changelog

This hotfix version contain two bug fixes:
1 - Ability to add guardrail from the agent builder section

• Adding Context ID to agent interaction logging: Developers embedding the SUPERWISE agent can now use the new Context ID parameter to log users’ interactions according to their relevant context (e.g., adding the client ID + date as context). This allows all interactions to be recorded with their associated context, making it easier to track and monitor interactions later under a given context. When using our iframe client, the iframe will automatically create a new context ID for any user session or when the same user hasn’t interacted with the agent for the last 10 minutes. If a Context ID is not supplied, it will be randomly generated for each interaction.
• Switching agent’s dataset: Users utilizing our observability capabilities for their agents can now choose to switch the dataset they log to. With this feature, users can log data from multiple agents into the same dataset, providing an easier way to perform cross-agent queries and policies.
• Tags: To better enable users to organize their agents flexibly, we now allow multiple tags to be added to all resources (agents, policies, dashboards, datasets, guardrails). Tags are system-level, so users can reuse the same tag (e.g., use-case-oriented) for relevant guardrails, agents, or any other resource. These tags make search and filter operations simple and efficient.
• Pricing tiering changes: To simplify the SaaS tiering logic for users, we have created a limit per resource per tier. Each tier comes with independent limits for the number of agents, policies, guardrails, datasets, dashboards, and users, instead of the agent + dataset combination we had previously.
• Additional improvements: A few UX bug fixes, fixes for MCP server providers such as Composio, implementation of marketing tracking tags, and framework upgrades.

• Overall web framework upgrade - By using the latest version of our web framework, we’ve achieved better web performance and dramatically reduced the number of vulnerabilities in our platform.
• Remote MCP Integration - With this new option, users who build “Advanced Agents” can now add a remote MCP tool. MCP is the new standard for exposing capabilities in a way that agents can easily consume to extend their native capabilities, context, and data source. Integrating an MCP server is powerful, as it allows our users to leverage the vast array of publicly available MCP servers. With this integration, users can potentially connect to any data source via the relevant MCP server. Remote MCP assumes that the end user wants to leverage an existing (or their own custom-made) MCP server already deployed elsewhere. Our integration supports a flexible way to configure the needed header parameters to integrate with various authentication methods.
• Delete Guardrail - To complete our previously deployed guardrail capability, we now allow users to completely delete a guardrail set of rules. Even when deleted, the configuration history of connected agents will show what was connected in the past.
• Teams Integration on Signup - To support our user acquisition efforts, any time a new tenant is created (a.k.a. signup), a notification will be made in the “sw-new-tenant” channel in Teams (previously, this was only on Slack).
• Bug fix - This version contains a fix for the table view displaying resources correctly across different resolutions.

This hotfix release fixes the colors of each business tier as viewed in the platform.

This is a major milestone as this version contains the following significant updates:

This release introduce the following improvements:

This hotfix contains two important bug fixes:

New Agent Studio Experience ✨
Our new agent management studio includes three main parts:

This release contains two important fixes:

This hotfix version adds more logs to detect the root cause of specific edge cases that are causing timeouts and errors when querying our internal data warehouse