robinwintertaylor

Prompt Router — AI skill for Claude Code

AI community

Prompt-Router: lightning-fast LLM gateway with a telemetry dashboard for coding agents (Goose, Cursor, Claude Code, VS Code) and enterprise pipelines.

How to install Prompt Router

This entry records only its repository, not the path inside it, so there is no exact command to give. Open robinwintertaylor/Prompt-Router and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What Prompt Router does

Prompt-Router: lightning-fast LLM gateway with a telemetry dashboard for coding agents (Goose, Cursor, Claude Code, VS Code) and enterprise pipelines. Why pay frontier prices for trivial lookups? TypeSafe Jev System One sorts each prompt in under 120ms and sends it to the right model across Azure AI Foundry, Mammouth AI and OpenRouter. Wallets win.

Alternatives in AI

PyTorch Lightning — Deep learning with PyTorch Lightning framework 6.4k ★
Pilotfish — Multi-model orchestration layer for Claude Code — the frontier model plans, cheaper models execute, verificati 688 ★
Iran MCP — 39 MCP servers (488 tools) that give AI agents the Iranian web: Digikala, Torob, Divar, Snappfood, Jabama, MrB 151 ★

README

⚡ The LLM Router That Knows When Not to Switch ⚡

Schema-constrained routing across 540+ models using TypeSafe Jev System One.
Drop-in replacement for OpenAI API endpoints with real-time HUD optics and token telemetry.

🌀 Why Naive LLM Routers Break Coding Agents

Most LLM routers evaluate every prompt in isolation. For single-turn chat, that works. **For coding agents (Goose, Cursor, VS Code Continue), it is economically broken.**

Coding agents accumulate massive multi-turn conversation threads (15k to 100k+ tokens) containing repository maps, tool outputs, and code diffs. Modern frontier providers offer **75% to 90% prompt caching discounts**:

Anthropic: $0.30/M cached input vs $3.00/M uncached (90% discount)
DeepSeek: $0.0