Retrieve the triple preset configuration for CODE-MAS, detailing planner, coder, and reviewer roles.
MIT
521,454 tools. Updated 2026-09-06 11:42
"Self-Training Coder Resources and Tools" matching MCP tools:
- Analyzes a task and selects a suitable model preset from options: smart, cheap, creative, fast, and coder.-
- Launch training of a self-supervised encoder for a specific sulcal region, using dataset configurations created by start_config.-
- Generate code from natural language descriptions in specified programming languages using the Qwen3-Coder model.-
- Plan risk-adjusted action sequences over a receding horizon using a learned transition model. Accepts caller-supplied actions and returns a plan without self-training.-
- When you need to know what AI Dimag can do, get an overview of its tools, prompts, resources, and CLI commands.MIT
Matching MCP Servers
- AlicenseNot gradedqualityFmaintenanceLLM routing proxy that cuts API costs 60-90% by auto-selecting the cheapest capable model across OpenAI, Anthropic, and Google. Provides stats, config, and model comparison tools via MCP.211MIT
- AlicenseAqualityBmaintenanceA stateless MCP server that exposes tools for evidence-based strength training, including classical powerlifting programs and fatigue management.8MIT
Matching MCP Connectors
Train portable RVC v2 voice models from audio in the cloud and download the .pth, .index, and ZIP.
Engineering log of self-hosted AI on NVIDIA DGX Spark (GB10/SM121A). 60+ articles indexed.
- Connect to a nomos controller by name to access its tools and resources, automatically disconnecting any previously connected controller.MIT
- Discover available Velixar tools, resources, prompts, and features to see which cognitive capabilities are enabled.MIT
- Self-check for active drift, loops, stuck tools, and critical signals before a major task phase to prevent repeating mistakes.Business Source 1.1
- Search for trusted tools, articles, and resources recommended by domain experts, with explanations of their value.MIT
- Check MCP server health by retrieving adapter availability, loaded tools, resources, prompts, and recent successful calls.MIT
- Initiate a tracked training job on an immutable dataset version. Provide org and dataset slugs to start training, with optional compute, epochs, and preset.Apache 2.0
- Retrieve source-cited daily health, training, and nutrition facts to assess an athlete's recovery and load. Requires athlete ID.AGPL 3.0
- Download specific training datasets from CrowdCent prediction challenges to a specified .parquet file path for model training and analysis.MIT
- Manage ML training jobs: check prerequisites, start LoRA fine-tuning, monitor progress, cancel runs, list adapters, and view statistics.MIT
- List model versions with training status and EDA report readiness to track model progress.MIT
- Retrieve ML training pipeline runs to inspect status, parameters, and duration for effective monitoring and troubleshooting.Apache 2.0
- Retrieve available training datasets for CrowdCent prediction challenges to prepare data for model development and analysis.MIT
- Check system health and self-heal DB integrity, tools, memory, and semaphores. Restart or clean up stale caches to recover from issues, with dry-run for safe cleanup.MIT
- Consolidate sleep, HRV, Body Battery, readiness, training load, and resting heart rate into a single daily health snapshot, so you can review recovery and training metrics at a glance without multiple calls.MIT
- Get weekly TSS targets, training phases, and week notes from your annual training plan. Narrow the date range for a specific month or week.MIT