Skip to main content

ProActions Hub Release Notes

1.2.0 - 05/02/2026​

New Features​

Admin Panel​

A new built-in browser-based admin panel for monitoring and diagnostics.

  • Secure authentication — stateless API key authentication (auth.admin.apiKeys) independent of the main auth system; minimum 32-character key requirement; optional IP/CIDR allowlist via auth.admin.allowedNetworks
  • Dashboard — process information, memory usage, configuration summary, and runtime usage charts
  • AI-Link metrics — real-time token usage per provider/model, latency distributions, finish reason breakdowns, and per-action request tables
  • Proxy usage — configured proxies and per-proxy request counts
  • MCP monitoring — connection status, per-server latency/error metrics, and per-tool invocation counts
  • Config viewer — full runtime configuration with sensitive fields automatically redacted
  • System tab — Node.js version, uptime, and memory breakdown
  • Toolbar — auto-refresh (30 s), manual refresh, statistics reset with confirmation, dark/light mode toggle

See Admin Panel for full configuration and API reference.

AI-Link - SSE Streaming Support​

The AI-Link gateway now natively proxies Server-Sent Events (SSE) streaming responses.

  • Set "stream": true in the request body to receive a streaming response
  • The Hub forwards the upstream SSE stream directly to the client with proper text/event-stream headers
  • Client disconnects are detected and upstream connections are torn down cleanly to prevent resource leaks
  • Streaming requests are tracked separately in the Admin Panel (streaming ratio metric)
MCP Gateway - Remote Server Support​

  • HTTP Transport (transport: http):

    • Connect to remote MCP servers via Streamable HTTP protocol
    • Uses HTTP POST for sending requests and Server-Sent Events (SSE) for receiving responses
    • Custom headers support for authentication (e.g., Authorization: Bearer ...)
    • Ideal for centralized MCP servers or third-party hosted services

  • WebSocket Transport (transport: websocket):

    • Connect to remote MCP servers via WebSocket for low-latency, bidirectional communication
    • Suitable for real-time MCP servers requiring persistent connections

  • Backward Compatible:

    • stdio transport remains the default for local subprocess-based MCP servers
    • Existing configurations continue to work without changes
MCP Gateway - Connection Stability Improvements​

  • Configurable Timeouts:

    • connectTimeout: Connection timeout in milliseconds (default: 30000ms)
    • toolTimeout: Tool invocation timeout in milliseconds (default: 60000ms)

  • Connection Reliability:

    • Automatic transport cleanup on connection failure (prevents orphaned processes/connections)
    • Disconnect timeout protection during shutdown (prevents hanging on unresponsive servers)
    • Async connection drop detection via transport lifecycle handlers
    • Timer leak fixes in timeout handling
MCP Gateway - Server Log Routing​

stderr output from stdio MCP server subprocesses is now captured and routed through the Hub's structured logger instead of spilling directly to the parent process console. Each line is emitted at warn level with the server name and stream: stderr metadata, making MCP server diagnostic output visible in your standard log pipeline (file rotation, JSON export, etc.).

Documentation​

  • Updated MCP Gateway documentation with remote transport configuration examples
  • Added transport type comparison table
  • Added security considerations for remote connections
  • Added Admin Panel documentation
  • Added streaming support to AI-Link documentation

1.1.0 - 11/11/2025​

New Features​

Enhanced AI Provider Support​
  • 9 New AI Providers Added:
    • LEPTON - Lepton AI cloud platform
    • KLUSTER_AI - Kluster AI services
    • NSCALE - nScale AI infrastructure
    • HYPERBOLIC - Hyperbolic AI platform
    • BYTEZ - Bytez AI services
    • FEATHERLESS_AI - Featherless AI
    • KRUTRIM - Krutrim AI
    • QDRANT - Qdrant vector database AI
    • THREE_ZERO_TWO_AI - 302.AI platform
  • Added support for advanced configuration objects
  • Enables custom load balancing strategies and fallback configurations
  • Allows guardrail integration for content moderation
  • Provides flexibility for complex multi-provider setups
Custom OpenAI-Compatible Providers:​
  • Integration support for any OpenAI-compatible API endpoint
  • Configurable endpoint URL with placeholder support (e.g., {path})
  • Custom header management for provider-specific authentication

AI Gateway Reliability & Performance​

  • Gateway Readiness Checks:

    • Requests are now rejected with HTTP 503 if AI Gateway is not ready
    • Prevents request failures during gateway startup
    • Improves overall system reliability and user experience

  • Vertex AI Token Caching:

    • Implemented intelligent token caching for Google Vertex AI
    • Reduces authentication overhead and improves response times
    • Automatic token refresh and lifecycle management

  • AI Request Timeout Configuration:

    • Configurable timeout for AI requests to prevent hanging connections
    • Default timeout protection against unresponsive upstream services
    • Enhanced error details for timeout scenarios

Security Enhancements​

  • Security Scanning Toolchain:
    • Integrated comprehensive security scanning tools (Trivy, Gitleaks, Docker Scout)
    • Added npm scripts for vulnerability scanning workflows
    • Automated secret detection and CVE scanning
    • SBOM (Software Bill of Materials) generation support
  • Enhanced Session Cookie Security:
    • Consolidated cookie configuration under unified cookie key
    • Added security defaults: httpOnly, secure, sameSite options
    • Improved cookie secret management with automatic fallback and warnings
    • Better protection against CSRF and session hijacking attacks
  • Configuration Security:
    • Added validation options for stricter configuration checking
    • Improved configuration defaults for production deployments
    • Enhanced error messages for misconfigured security settings

Health Check Improvements​

  • Health Check Guard:
    • Added guard protection for health check endpoints
    • Configurable access control for monitoring endpoints
    • Supports deployment behind load balancers and orchestrators

Improvements​

Error Handling & Diagnostics​

  • Enriched HTTP Error Details:

    • Enhanced error messages with upstream connection details
    • Better diagnostic information for debugging AI provider issues
    • Improved logging of request/response errors with context

  • Improved Error Logging:

    • More detailed error traces for troubleshooting
    • Better filtering of sensitive data in error logs
    • Enhanced log filtering for circular references and complex objects

Logging Enhancements​

  • Improved Sensitive Data Filtering:

    • Enhanced detection of API keys, tokens, and passwords
    • Handles JSON-embedded credentials and circular object references
    • More robust redaction across all log types

  • Optimized Debug Logging:

    • Reduced verbose debug output in production
    • Cleaned up unnecessary debug logs in tools and YouTube controllers
    • Better performance with selective logging

Prime Panel Updates​

Status Indicator & Notifications​

Visual feedback system for communicating process status and notifications to users.

API Functions:​
  • window.ProActionsPanel.setStatus(statusText) - Displays a status bar at the bottom of the panel with animated indicators to show ongoing processes
  • window.ProActionsPanel.clearStatus() - Hides the status bar
  • window.ProActionsPanel.showNotification(message, duration?) - Shows a temporary toast notification (default: 3 seconds)

Infrastructure​

  • Node.js 22 Runtime:
    • Continued use of Node.js 22 for optimal performance
    • Updated Red Hat UBI8 base image with Node.js 22
    • Enhanced stability and security with latest LTS features

1.0.9 - 02/07/2025​

Maintenance​

Updated ProActions Core and AI-Kit to version 1.0.12 for Prime Panel.

1.0.8 - 30/06/2025​

Improvements​

YouTube API Integration​
  • Added optional thumbnail upload to YouTube Video upload.
Maintenance​
  • Updated libraries and dependencies.

1.0.7 - 28/04/2025​

We're excited to announce the release of ProActions Hub 1.0.7, featuring significant new capabilities and improvements to enhance your workflow integration with external services.

New Features​

YouTube API Integration​
  • Added comprehensive YouTube API support for authentication and video upload
  • Enables seamless content publishing directly to YouTube from within Swing workflows
  • Supports standard OAuth authentication flow for secure access to YouTube accounts
Prime Panel (Beta)​
  • Introduced beta support for the new Prime Panel
  • Allows ProActions workflows to be used directly from the Prime interface
JSON Logging​
  • Implemented structured JSON logging for improved log analysis and monitoring
  • Enhances debugging capabilities and system observability
  • This allows a direct integration with Méthode Logging Platform
TLS Encryption Support​
  • Added native TLS/SSL encryption for secure communication
  • Ensures data transmitted between ProActions Hub and external services remains private
  • Configurable through the config.yaml configuration file
Swagger Documentation​
  • Integrated Swagger UI for comprehensive API documentation
  • Provides interactive exploration of ProActions Hub API endpoints
Expanded LLM Provider Support​
  • Added support for five new LLM/AI service providers:
    • Cortex
    • Nebius
    • Recraft-AI
    • Milvus
    • Replicate

Improvements​

  • Added the ability to customize the cookie name used for authentication

1.0.6 - 13/12/2024​

New Features​

  • Support for Additional AI Services:
    • xAI Grok API: Added integration for xAI Grok, expanding the range of supported AI services.
    • Lambda Labs: Enabled support for workflows using Lambda Labs.
    • DashScope: Integrated DashScope, allowing seamless interaction with its capabilities.
    • SageMaker: Added support for AWS SageMaker, enhancing machine learning workflow options.

Improvements​

  • Enhanced Error Handling:
    • Improved handling of remote service errors by including the original error details in response messages. This provides better transparency and debugging information.

Maintenance​

  • Docker Image Update:
    • Updated the Docker image to use Node.js v20.

1.0.4 - 01/10/2024​

  • Support for additional providers
    • azure-ai
    • huggingface
    • deepseek
    • predibase
    • triton
    • voyage
    • github
    • deepbricks
    • siliconflow
    • cerebras
    • inference-net
    • sambanova
    • upstage

1.0.3 - 30/09/2024​

  • Fixed http communication issue with invalid responses.

1.0.2 - 20/09/2024​

  • Added private key authentication for Google Vertex AI.