Multi-LLM Orchestration Router for Mac Studio
Inspired by Sakana AI's Fugu (which matches Fable 5 benchmarks by dynamically routing tasks to the best model in a pool), build a local orchestration layer for the Mac Studio that routes agent tasks across Ollama models, Claude API, and local alternatives.
- ▸router/router.py — Core routing engine: classify → privacy gate → memory gate → dispatch
- ▸router/classifier.py — 8-rule heuristic classifier maps prompts to tiers in <1ms
- ▸router/memory_monitor.py — macOS memory pressure probe with 30-second cache
- ▸router/config.py — Pydantic config loader with actionable validation error messages
- ▸router/providers/ — Adapters for Ollama, LM Studio (OpenAI-compat), and Anthropic SDK
- ▸config.yaml — User-editable tier assignments, model names, and provider URLs
- ▸tests/test_router.py — 12-test verification suite (classifier, privacy gate, memory gate, lazy API key init)
- ▸guide/SETUP.md — First-run walkthrough: install, configure, first prompt
- ▸guide/HOW-IT-WORKS.md — Design rationale, routing pipeline diagram, extension guide
- ▸guide/ROUTING-REFERENCE.md — Quick-reference: tier table, classifier rules, full CLI and Python API
## goes-well-with
From the same shelf
Agentic AI Security Policy Template Pack
Downloadable Markdown + YAML policy templates giving IT teams the acceptable-use, blast-radius, approval-gate, and logging scaffolding to safely deploy agentic AI coding tools like Claude Code, Codex, and Grok Build.
Agentic Migration Playbook: API Endpoint Modernization
For DevOps and platform engineers with a real migration deadline — the complete Claude Code workflow you'd otherwise spend months reverse-engineering.
Ansible Automation Starter Kit for Linux Sysadmins
12 production Ansible playbooks for Linux sysadmins — annotated, tested, ready to run.
## not-ready-to-buy
Take the field notes instead
One practical write-up a week from the same workbench these kits come from — plus reader pricing when new kits ship.