2.4.0.3 release notes

2.4.0.3

• Issue 2919 (redux) and Issue 2955 - Improved handling of references to *-meta.xml files when those files are renamed, particularly in other XML files such as package.xml and permission sets/profiles.
• Issue 2949 - Prospective fix for an issue where loaded scratch org definition files in the create scratch org dialog could be based on stale cached information.
• Issue 2952 - Improved Apex offline debugger source position determination when stub/synthetic declarations are involved, e.g., default no-arg constructors.
• Issue 2956 - Fixed an issue where Apex coverage would not be displayed properly after unit test execution.
• Extended IC's assisted installation/configuration support for agentic AI tools to include more agentic AI JetBrains plugins and more/more recent Salesforce-specific support components. A demonstration video is available here.
  
  In Illuminated Cloud | Configure Application | Agentforce, the Agentforce Vibes section has been renamed Vibe Coding to reflect this more general support for agentic AI tools, and it now includes a tabbed interface with support for:
  • Claude Code
  • Cline
  • GitHub Copilot
  • JetBrains AI Assistant and Junie
  • Kilo Code
  • Windsurf
  • Zencoder
  IC assists with installation and configuration of each of these agentic AI tools including:
  • Plugin installation/enablement
  • Salesforce DX MCP server configuration
  • Salesforce hosted MCP servers configuration - See below for additional details.
  • Salesforce-specific rules - Note that the bundled rules have been updated significantly vs. earlier versions of this integration to include those that guide the tools toward proper and consistent usage of the Salesforce MCP servers and bundled skills.
  • Salesforce-specific skills
  In general IC tries to add project-independent configuration at the user-level (i.e., under the user home directory) and project-specific configuration at the project-level (i.e., under the project root directory). However, based on each tool's own preferences and limitations, there are some tool-specific variances. IC shows the configuration scope of each tool/configuration combination, and once installed, the scope acts as a link to the actual configuration file(s).
  
  Salesforce hosted MCP servers considerations
  The Salesforce hosted MCP servers generally require configuration of an External Client App (ECA) for MCP client authentication and authorization. In order to simplify these turnkey tool integrations, IC provides an option to Authorize Salesforce hosted MCP server requests automatically using its configured Salesforce org connections. This option is disabled by default, but when enabled, IC automatically queries the active MCP servers from the connected organization — and provides a convenient link to configure the org's active MCP servers if necessary — and configures the agentic AI tools to proxy all Salesforce hosted MCP server requests through IC so that those requests can be automatically decorated with IC's authorization header. Note that this proxy behavior does not inspect, capture, or otherwise compromise the contents of the proxied requests and responses, and all communications with the Salesforce hosted MCP servers are, of course, performed over HTTPS/SSL.
• Many other fixes and improvements.

Other news

• The next major task will be adding full, first-class support for the new UIBundle metadata type and the associated on-platform support for React. Again, React is also now supported by the free JetBrains 2026.1 IDEs, so once this feature has been delivered, all IC users should have access to a quite complete set of React/Salesforce development tools.
• I realize this is a repeat, but in the spirit of overcommunicating important things, after that I'm planning to begin moving the public issue tracker and IC user guide from Bitbucket to GitHub in earnest. If you've visited either of those recently, you've no doubt seen the header banner about those being shut down on August 20, 2026, but I plan to have things migrated well in advance of that date. Feel free to continue using them in their current locations between now and the move, though, as they continue to be fully operational in the interim. As before, if you'd like your issue tracker activity reflected in the new issue tracker, please provide your Bitbucket and GitHub user IDs so that I can include them in the user mapping.