Correct a wrong model routing by redirecting to a better tool and logging the correction to improve future routing decisions.
MIT
457,697 tools. Updated 2026-08-14 14:03
"LLVM" matching MCP tools:
- Run tasks on Google's Gemini models via local CLI, using your Google One AI Pro subscription. Use as a fallback when Claude limits are tight or for tasks benefiting from Gemini.MIT
- Analyze workspace files to generate a compact semantic summary, giving LLM routing the project context it needs to choose the right model.MIT
- Get project sequence details (name, prefix, pad, last_used, next, next claim ID) to avoid guessing conventions; ask before seeding missing sequences.AGPL 3.0
- Recommends the cheapest LLM model fitting your token workload and budget, with optional provider and model filters.MIT
- Auto-route AI prompts with persistent server-side savings tracking, ensuring accurate cumulative value even on hosts lacking client-side hooks.MIT
Matching MCP Servers
- AlicenseAqualityCmaintenanceAn MCP server that runs curated adversarial prompts against local Ollama models to test guardrails, scoring responses with heuristic verdicts for human review.5MIT
- FlicenseAqualityBmaintenanceMCP server to help AI assistants generate Italian PA user interfaces by providing components, recipes, accessibility rules, and validation for Bootstrap Italia and Web Components.9
Matching MCP Connectors
Forecast future events and scan prediction-market edges.
Swiss weather data for AI assistants — forecasts, measurements, stations, pollen.
- Resolve blocked high-cost AI routing calls: approve to proceed, cancel, or downgrade to a cheaper model.MIT
- Monitor real-time budget pressure for all configured AI providers. See which are fully available or exhausted to plan cost-efficient routing.MIT
- Rate AI routing decisions as good or bad to store feedback, improving future model selection.MIT
- List all participants in the project roster with their name, kind, lane, admin status, and token activity. Use this to check for duplicates before provisioning a new admin.AGPL 3.0
- Convert images from local files, URLs, clipboard, or base64 into detailed text descriptions, giving text-only agents vision capabilities.MIT
- Classifies task complexity to route prompts to the optimal LLM tier—budget, balanced, or premium—saving cost while maintaining quality.MIT
- Evaluate and benchmark local and remote models on reasoning and code tasks to measure quality, speed, and accuracy for optimal routing.MIT
- Route AI tasks to the local Codex desktop agent via Codex CLI for non-interactive execution using your OpenAI subscription. Choose it when Claude limits are tight or tasks fit OpenAI's models.MIT
- Generate bulk edit instructions across multiple files for cross-file refactors, bulk renames, and updating repeated patterns in one pass.MIT
- Query LLM spending grouped by provider, model, project, tag, or day over a customizable time window, with optional filters to narrow results.MIT
- Compare LLM providers and models by projected cost for a given workload. Ranked by absolute cost, with relative cost percentages against the cheapest option.MIT
- Assess a prompt's complexity to get a model recommendation that weighs budget, quality, and minimum model tier, and includes a budget usage bar.MIT
- Classify your task prompt to get the recommended AI agent CLI and model for session-level routing. Matches task complexity and budget to the best Claude, Codex, or Gemini agent.MIT
- Send a query to the best available LLM. Automatically routes by complexity—simple to cost-effective models, moderate to balanced, complex to premium—with optional controls for model, temperature, system prompt, and context.MIT