MCP Media Engine_
// CHANGELOG

Track our progress.

Every update, feature, and fix — documented with version history. Stay informed about what's new in MCP Media Engine.

Latest: v1.1.0
v1.1.0

Prompt Library, Remote MCP, and Distribution Everywhere

MediaEngine is now reachable from every major MCP directory and any remote MCP client, ships a copy-paste prompt library generated with its own API, and publishes honest engineering guidance from real production testing.

Added
  • •Prompt library at /prompts — 10 copy-paste prompts across selfies, pets, product shots, spaces, and style art, each with a real generated example and its job ID as receipt
  • •Hosted remote MCP endpoint at mcp.mcpmediaengine.com/mcp — connect from claude.ai, ChatGPT, or any streamable-HTTP MCP client without installing anything
  • •Listed on the official MCP Registry, mcp.so, Smithery, and Glama — one-click discovery from every major MCP directory
  • •Model pages at /models with per-model guides, and six new engineering blog posts covering agent image-gen setup, async pipeline patterns, img2img style-transform testing, and cost math
Changed
  • •Dashboard now shows live job history and usage analytics backed by new session-authenticated API routes
  • •img2img style transforms documented with honest guidance: watercolor and cartoon transforms are reliable at denoising 0.75–0.85; material transforms (clay, LEGO-style) are not — verify visually before publishing
v1.0.2

Production API Launch with Billing & Model Catalog

The production API is fully launched: usage-based billing, a real model catalog with per-model costs and aspect ratios, and docs that show an agent-specific setup snippet for every major MCP client. Free tier (25 credits) remains on the first API key.

Added
  • •Aspect-ratio support on image and video jobs (1:1, 16:9, 9:16, 4:3, 3:4) — pick the right shape per channel without manual cropping
  • •Model catalog now lists per-model credit cost, supported aspect ratios, and average latency so you can budget jobs before you submit
  • •Documentation hub at /docs ships a 'Works with your agent' grid covering Claude Desktop, Claude Code, Cursor, Windsurf, Cline, Codex CLI, Gemini CLI, Goose, generic MCP, and REST API
  • •llms.txt at /llms.txt refreshed with the agent compatibility list, per-agent MCP config snippets, and the v1.0+ billing/free-tier guidance
Changed
  • •Truth Gate: removed the placeholder link to a phantom modelcontextprotocol/servers directory; the canonical MCP server lives at npmjs.com/package/mcp-media-engine
v1.0.1

Stability and Hardening

A reliability and hardening pass: a more stable dashboard, smoother signup, correct media content types, and clearer error messages.

Fixed
  • •Dashboard home no longer errors while loading the usage analytics panel
  • •More reliable signup and first-key free-tier credit grant
  • •Media job output files now serve with correct image/video content types
Changed
  • •Authentication and reliability hardening across the API
  • •Clearer, friendlier error messages on failed jobs
v1.0.0

General Availability — the API is live

MediaEngine is generally available: sign up, get an API key, and generate hero images and teaser videos through a single async API call. Usage-based billing, a free tier to start, and a live dashboard.

Added
  • •Public API live at api.mcpmediaengine.com — self-serve signup, instant API keys, and one async call from content to finished media
  • •Self-serve billing via Stripe Checkout with usage-based credits
  • •Free tier: 25 credits on your first API key, no card required
  • •Model catalog endpoint (GET /api/models) with per-model credit costs — Nano Banana, GPT Image 2.0, Seedance, and Hailuo
  • •Dashboard at app.mcpmediaengine.com — jobs, API keys, usage, and billing in one place
Changed
  • •Pricing aligned to real backend delivery — honest credit allocations, no over-promised monthly numbers
Fixed
  • •Reliability and security hardening across the API and authentication layer
v0.9.0

Batch Processing and Webhook Improvements

MCP Media Engine now supports batch processing with up to 500 jobs per request, improved webhook reliability with retry logic, and real-time job status streaming.

Added
  • •Batch mode supports up to 500 jobs per request
  • •Webhook retry logic with exponential backoff
  • •Job status streaming via Server-Sent Events (SSE)
Changed
  • •Improved queue latency to under 2 seconds average
  • •Enhanced error messages for invalid prompts
Fixed
  • •Fixed race condition in webhook delivery
  • •Resolved memory leak in long-running job monitoring
v0.8.0

MCP Protocol v1.1 Compatibility

Full compatibility with MCP Protocol v1.1, enabling native integration with Claude Desktop and Cursor agent workflows.

Added
  • •Native MCP v1.1 tool definitions
  • •Claude Desktop integration guide
  • •Cursor agent MCP configuration
Changed
  • •Updated stdio transport to support v1.1 protocol
  • •Enhanced resource templates for job results
Fixed
  • •Fixed MCP authentication token refresh
  • •Resolved connection stability issues in long sessions
v0.7.0

Video Generation and Audio Modes

Video and audio generation modes added, expanding the platform beyond image generation to support full media workflows.

Added
  • •Video generation models
  • •Audio generation with multiple voice options
  • •Article teaser video generation tool
Changed
  • •Expanded the model catalog
  • •Updated pricing page with per-mode credits
Deprecated
  • •Legacy synchronous generation endpoint (use async instead)
v0.6.0

Credit-Based Billing System

Implemented credit-based billing, making it easier to manage and track media generation costs across teams and projects.

Added
  • •Credit-based billing with usage dashboards
  • •Team member credit allocation
  • •Credit purchase and top-up system
Changed
  • •Replaced per-request pricing with credit consumption model
  • •Updated API rate limits per plan tier
v0.5.0

Image Generation v2

Major update to image generation with an upgraded pipeline and improved output quality.

Added
  • •High-quality image generation support
  • •Video-to-image generation pipeline
  • •Image style presets and guidance controls
Changed
  • •Improved image quality with v2 upscaling
  • •Faster generation times with optimized queue
Fixed
  • •Fixed aspect ratio handling for non-standard sizes
  • •Resolved prompt truncation for long descriptions
v0.4.0

Webhook System and Reliability

Production-grade webhook system with monitoring, retry logic, and guaranteed delivery for mission-critical applications.

Added
  • •Configurable webhook endpoints per job
  • •Webhook delivery status monitoring
  • •Dead-letter queue for failed deliveries
Changed
  • •Implemented 99.9% uptime SLA
  • •Added health check endpoints with Prometheus metrics
Fixed
  • •Fixed webhook timeout handling
  • •Resolved job state inconsistency under high load
v0.3.0

MCP Server Launch

Launched the MCP server, enabling direct integration with AI agents through the Model Context Protocol.

Added
  • •MCP server implementation with stdio transport
  • •HTTP/SSE transport for remote deployments
  • •Tool definitions for image generation workflows
Changed
  • •Migrated from REST-only to MCP-native architecture
  • •Updated authentication to support API key per workspace
v0.2.0

Core API and Queue Infrastructure

Core API infrastructure with async job queuing, enabling reliable media generation at scale.

Added
  • •Async job queue with BullMQ
  • •Job status polling and retrieval
  • •List jobs with filtering and pagination
Changed
  • •PostgreSQL for job persistence
  • •Redis for queue management
Fixed
  • •Fixed job priority ordering
  • •Resolved connection pool exhaustion under load
v0.1.0

Initial Release

Initial release with image generation and basic deployment infrastructure.

Added
  • •Image generation
  • •Basic API authentication
  • •Docker container deployment
  • •Cloud hosting infrastructure
// FAQ

Common questions.

What is MCP Media Engine?

MCP Media Engine is an async API for generating images, video, and audio. It uses an async job queue pattern where you submit a generation job and receive the result via webhook when complete. This differs from synchronous APIs that require polling or blocking requests.

How do I track changes to the API?

The changelog page documents every update, including new features, model additions, and bug fixes. Each release includes the version number, release date, and a summary of changes. You can also subscribe to updates via the RSS feed or check the /llms.txt for machine-readable documentation.

How often is MCP Media Engine updated?

We ship updates regularly with new features, new models, and improvements. Major releases happen roughly monthly, with smaller fixes deployed as needed. Subscribe to the changelog to stay informed about updates relevant to your integration.

How do I get notified of new features?

Webhook notifications let you receive real-time updates when jobs complete. For platform updates, subscribe to the RSS feed or check this changelog page periodically. Major changes are also documented in the API reference.

What is the difference between versions?

Each version number follows semantic versioning (MAJOR.MINOR.PATCH). Major versions may include breaking changes. Minor versions add features in a backward-compatible way. Patch versions include bug fixes and small improvements. We always provide migration paths for major changes.