Skip to content

Latest commit

 

History

History
255 lines (205 loc) · 10.6 KB

CHANGELOG.md

File metadata and controls

255 lines (205 loc) · 10.6 KB

Changelog

All notable changes to the Visionary Tool Server project will be documented in this file.

The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.

[Unreleased]

Added

  • Contributing guide with comprehensive development documentation
  • Changelog following Keep a Changelog format

[4.0.0] - 2026-02-06

Added

Core Infrastructure

  • 331 MCP tools across 64 modules providing comprehensive AI agent capabilities
  • FastMCP 2.14.2 framework for Model Context Protocol server implementation
  • Python 3.14.2 runtime with modern async/await architecture
  • Starlette 0.50.0 ASGI framework for high-performance HTTP handling
  • OAuth 2.1 with PKCE authentication system for secure API access
  • Self-test health check suite for automated tool validation and monitoring
  • Agent orchestrator with multi-agent chain execution capabilities
  • SSE (Server-Sent Events) transport on port 8082 for real-time communication

LLM Provider Coverage (13 Providers)

  • OpenAI: GPT-4, GPT-3.5, embeddings, fine-tuning, DALL-E integration
  • Anthropic: Claude 3 Opus/Sonnet/Haiku, extended context windows
  • Groq: Ultra-fast inference, Mixtral, Llama models
  • DeepSeek: Code-specialized models, competitive pricing
  • Mistral: Mixtral 8x7B, Mistral Large, European AI compliance
  • Grok: X.AI's conversational models
  • Perplexity: Search-augmented generation, real-time information
  • Cohere: Command models, embeddings, reranking
  • Together: Open-source model hosting, custom deployments
  • Fireworks: Optimized inference, function calling
  • OpenRouter: Unified API for 100+ models
  • HuggingFace: Open-source model hub integration
  • MiniMax M2.1: Chinese language optimization

DevOps & Project Management (90+ tools)

  • GitHub (37 tools): Repository management, issues, pull requests, actions, releases, webhooks
  • GitLab (26 tools): Project management, CI/CD pipelines, merge requests, wiki integration
  • Jira: Issue tracking, sprint planning, workflow automation, Jira Query Language (JQL)
  • Linear: Modern issue tracking, project workflows, cycle management
  • ClickUp: Task management, time tracking, goals, custom fields
  • Asana: Project planning, task dependencies, portfolio management
  • Monday: Visual project boards, workflow automation, team collaboration

Infrastructure & Deployment (15+ tools)

  • Vercel: Serverless deployment, preview environments, edge functions, analytics
  • Railway: Container deployment, PostgreSQL/Redis provisioning, environment management
  • Cloudflare: DNS management, edge workers, caching, DDoS protection, SSL/TLS

Monitoring & Observability (30+ tools)

  • Datadog: Full-stack monitoring, APM, log aggregation, custom metrics, dashboards
  • Prometheus: Metrics collection, alerting, PromQL queries, service discovery
  • Grafana: Visualization, dashboard creation, alert management, data source integration
  • Sentry: Error tracking, performance monitoring, release tracking, issue grouping

AI Media Generation (25+ tools)

  • ElevenLabs: Text-to-speech, voice cloning, multilingual support
  • Replicate: AI model inference, Stable Diffusion, SDXL, custom models
  • FAL: Fast AI inference, image generation, video processing
  • Stability AI: Stable Diffusion, image editing, upscaling, inpainting

Search & Web (15+ tools)

  • Firecrawl: Web scraping, content extraction, structured data
  • Exa: Semantic search, knowledge graph queries
  • Brave Search: Privacy-focused search, no-tracking API

Communication (35+ tools)

  • Discord (21 tools): Server management, channels, messages, roles, webhooks, embeds, reactions
  • Slack: Messaging, channels, threads, file sharing, app integration
  • Twilio: SMS, voice calls, WhatsApp, phone number management
  • SendGrid: Transactional email, templates, tracking, analytics
  • Resend: Modern email API, React Email templates, webhook handling

Payment Processing (10+ tools)

  • Stripe: Payment intents, subscriptions, customer management, invoicing, webhooks

Knowledge Management (12 tools)

  • Obsidian: Vault integration, note creation, linking, search, tag management, daily notes

Agentic Tools (15+ tools)

  • Nemotron: NVIDIA's agentic AI capabilities, tool chaining, context management
  • Agent Orchestrator: Multi-agent workflows, task delegation, result aggregation

vLLM Integration (22 tools)

  • Razer AIKit: Local LLM inference with RTX 4090 GPU acceleration
  • Model Management: Load, unload, list models, memory optimization
  • Fine-tuning: LoRA training, model adaptation, checkpoint management
  • Benchmarking: Performance testing, latency measurement, throughput analysis

Changed

  • Upgraded FastMCP from 1.x to 2.14.2 with improved SSE handling
  • Migrated to Python 3.14.2 from 3.11 for performance improvements
  • Updated Starlette to 0.50.0 for enhanced ASGI support
  • Improved error handling with specific exception types and detailed messages
  • Enhanced logging with structured context and request tracing
  • Optimized Docker images for faster builds and smaller size

Security

  • OAuth 2.1 with PKCE replacing basic token authentication
  • Environment variable validation preventing insecure configurations
  • API key rotation support for zero-downtime credential updates
  • Rate limiting per tool and per API key
  • Request sanitization preventing injection attacks

Performance

  • Async/await throughout for non-blocking I/O operations
  • Connection pooling for external API calls
  • Response caching for frequently accessed data
  • Streaming responses for large payloads
  • GPU acceleration for local LLM inference

[2.0.0] - 2026-01-09

Added

Razer AIKit vLLM Integration

  • 22 vLLM tools for local large language model inference
  • RTX 4090 GPU support with CUDA optimization
  • Model management: Load, unload, switch between models dynamically
  • Fine-tuning capabilities: LoRA, QLoRA training on custom datasets
  • Benchmarking suite: Performance metrics, latency analysis, throughput testing
  • Memory optimization: KV cache management, quantization (4-bit, 8-bit)

Docker Production Stack

  • Docker Compose configuration for production deployment
  • Multi-stage builds for optimized image size
  • Health checks for container orchestration
  • Volume mounting for model persistence
  • Environment-based configuration for dev/staging/prod

Local LLM Tools

  • vllm_load_model: Load models with configurable parameters
  • vllm_generate: Text generation with streaming
  • vllm_chat: Conversational interface
  • vllm_embeddings: Local embedding generation
  • vllm_fine_tune: Model fine-tuning workflows
  • vllm_benchmark: Performance testing
  • vllm_list_models: Available model enumeration
  • vllm_unload_model: Memory management

Changed

  • Enhanced GPU utilization with tensor parallelism
  • Improved model loading with automatic format detection
  • Optimized VRAM usage with PagedAttention

Performance

  • 50-100ms latency for local inference on RTX 4090
  • 2000+ tokens/sec throughput for Llama 2 7B
  • Multi-GPU support for larger models

[1.0.0] - 2025-12-01

Added

Initial Release

  • MCP server with FastMCP framework
  • SSE transport on port 8082
  • Basic health check endpoint at /health
  • Docker support with Dockerfile and docker-compose.yml

GitHub Integration (12 tools)

  • github_create_issue: Create issues with labels and assignees
  • github_list_issues: List and filter repository issues
  • github_get_issue: Get detailed issue information
  • github_create_pull_request: Create PRs with description
  • github_list_pull_requests: List and filter PRs
  • github_merge_pull_request: Merge PRs with options
  • github_create_repository: Create new repositories
  • github_list_repositories: List user/org repositories
  • github_create_branch: Create new branches
  • github_list_branches: List repository branches
  • github_get_commit: Get commit details
  • github_list_commits: List commit history

Discord Integration (8 tools)

  • discord_send_message: Send messages to channels
  • discord_create_channel: Create text/voice channels
  • discord_list_channels: List server channels
  • discord_create_role: Create roles with permissions
  • discord_list_roles: List server roles
  • discord_send_dm: Send direct messages
  • discord_create_webhook: Create channel webhooks
  • discord_list_webhooks: List channel webhooks

Core Features

  • Type-safe API: Full type hints and validation
  • Error handling: Structured error responses
  • Logging: Request/response logging with context
  • Configuration: Environment-based settings
  • Documentation: OpenAPI/Swagger auto-generated docs

Dependencies

  • fastmcp==1.0.0
  • starlette==0.45.0
  • uvicorn==0.30.0
  • pydantic==2.8.0
  • httpx==0.27.0
  • python-dotenv==1.0.0

Version History Summary

Version Release Date Highlights
4.0.0 2026-02-06 331 tools, 13 LLM providers, OAuth 2.1, agent orchestration
2.0.0 2026-01-09 vLLM integration, RTX 4090 support, local inference
1.0.0 2025-12-01 Initial release, GitHub + Discord, basic MCP server

Links

Migration Guides

Migrating from 3.x to 4.0.0

Breaking Changes:

  1. OAuth 2.1 Required: Basic token authentication removed
  2. New Environment Variables: Many _TOKEN variables renamed to _API_KEY
  3. Tool Naming: Some tools renamed for consistency

Steps:

  1. Update .env file with OAuth credentials
  2. Replace GITHUB_TOKEN with GITHUB_API_KEY
  3. Update tool calls to use new names (see documentation)
  4. Test thoroughly before deploying

Migrating from 1.x to 2.0.0

Breaking Changes:

  1. Python 3.14+: Minimum Python version increased
  2. FastMCP 2.x: Updated MCP protocol

Steps:

  1. Update Python to 3.14.2+
  2. Update dependencies: pip install -r requirements.txt
  3. Rebuild Docker images
  4. No tool API changes required

For detailed migration instructions, see MIGRATION.md

For contributing guidelines, see CONTRIBUTING.md