Changelog

Rate Limiting and Connections

• Added automatic rate-limit defaults for critical providers including Stripe, HubSpot, Mailgun, and Shopify to help prevent API bans
• Implemented validation to reject invalid rate-limit configurations
• Enhanced observability with metrics and alerts for fail-open scenarios
• Improved UI clarity with honest protection state badges

Report Performance

• Automatically hide large undisplayed columns in reports for faster rendering
• Optimized data fetching to load only displayed columns
• Introduced concurrent block rendering and schema caching
• Added batched chart aggregates to eliminate N+1 queries

Execution and Payload Handling

• Increased instance creation body limit to 64 MB
• Expanded backend file size support to 96 MB
• Automatically hide large input and output data by default with an option to view full content
• Added atomic instance claiming to prevent duplicate execution

Authoring and Tools

• Enabled tojson template helper with documented allowlist
• Added MCP tools for folder management including move, delete, and list
• Expanded authoring schema with full condition operator enum
• Implemented Python-style negative array index resolution

AI Agent

• Introduced AI Agent scenarios — build interactive, multi-step AI workflows that can reason, take actions, and respond to user input in real time
• Conversational chat interface — interact with running AI Agent scenarios through a real-time chat experience with streaming responses
• Human-in-the-loop support — AI Agents can pause execution and request input from users, with rich form rendering (labels, placeholders, validation, conditional visibility)
• Multi-turn sessions — conversations persist across multiple execution turns, allowing back-and-forth dialogue with an AI Agent without losing context
• Agent memory — AI Agents can load and save data to the object model, enabling persistent knowledge across conversations
• Structured output — AI Agents can return structured, validated responses in addition to free-form text
• XLSX spreadsheet parsing — new built-in agent that reads and extracts data from Excel spreadsheets

MCP (Model Context Protocol) Integration

• MCP server endpoint — expose your scenarios, executions, and object model as MCP tools, allowing AI assistants (like Claude, Cursor, etc.) to interact with Runtara programmatically
• MCP OAuth 2.1 support — standards-compliant authentication for MCP clients with automatic authorization server discovery (RFC 9728)
• Signal interaction via MCP — submit responses to human-in-the-loop steps through MCP tools, enabling fully automated agent-to-agent workflows

Scenario Builder

• Output schema editing — define and edit the output schema for your scenarios directly in the builder
• Template mapping improvements — better support for template values in step configuration with an improved mapping editor
• Connection category icons — connections now display category-specific icons for easier visual identification

Bug Fixes

• Fixed scenario recompilation delay — toggling debug mode no longer unnecessarily invalidates compiled scenarios, eliminating a ~5-7 second hang after compilation
• Security and performance improvements

• Problem tab now shows both server and client validation issues in a single view
• Fixed Problem tab filter switching not updating the displayed list
• Fixed crash when viewing error details with missing attributes
• Error steps now appear in execution history with full debug information
• Clear validation errors for scenarios with incomplete step configurations
• Execution status now shows the correct state instead of always displaying “in progress” in some cases
• Removed invalid statuses from the Invocation History filter dropdown
• HTTP agent connection dropdown now populates correctly
• Boolean fields in connection forms now render and save properly
• Improved Shopify private app support including client credentials authentication
• Fixed nested scenario parsing during compilation

• Added scenario import/export — back up and migrate scenario definitions between environments
• Added MCP (Model Context Protocol) integration for AI-assisted scenario management
• Added API key authentication support for programmatic access
• Improved rate limiting across the platform with better visibility and controls
• Improved handling of truncated events in execution history
• Improved OpenAI connection handling and error reporting
• Fixed LENGTH operator handling inside nested expressions

• Added calendar component for scheduling operations
• Improved folder structure and organization for scenarios
• Unified unsaved changes prompts in the scenario editor
• Fixed PWA update prompt not reloading correctly
• Improved scenario validation rules
• Added automatic cleanup of stale compiled scenario images

• Added Filter and GroupBy step outputs to the variable picker
• Added ID column header to object instance tables
• Added ability to open error configuration directly from error handles
• Improved schema name validation for database filtering
• Improved Error step field serialization
• Fixed incomplete filter conditions being sent to the backend
• Fixed child scenario default variables lost during compilation
• Improved connection stats and tagging in the execution engine
• Added variable display in instance history view

• Added WaitForSignal step — pause a scenario and resume it via an external signal
• Added Delay step for portable, timed waits within workflows
• Added child scenario input validation with circular dependency detection
• Added ensure-array capability for Split steps
• Improved single-instance trigger enforcement
• Improved rate limit API and visibility
• Improved heartbeat handling and process termination
• Streamlined execution platform internals for better reliability

• Added Switch step support in the visual editor
• Added GroupBy step for aggregating data within workflows
• Added Filter step for conditional data filtering
• Introduced OpenTelemetry-based observability (tracing, metrics, logs)
• Improved rate limit display and controls in the dashboard
• Improved cancellation and pause/resume handling
• Improved child scenario dependency tracking
• Added folder-specific scenario fetching

• Added bottom panel with history tab for quick access to execution events
• Added template editor for configuring step templates
• Introduced structured error handling with dedicated error step configuration
• Improved auto-layout algorithm for workflow graphs
• Improved edge rendering and scenario editing UX
• Added modular Shopify sync scenarios
• Improved cancellation handling to prevent deadlocks
• Added scope-specific filtering for execution history

• Added Progressive Web App (PWA) support — install the dashboard as a native app
• Added structured error processing with detailed error context
• Added Error step with configurable error handling paths
• Added connection-based object store support
• Improved condition editor UX
• Fixed SFTP connection editing
• Improved mobile responsiveness