Loading...
Sold by manteclaw with secure USDC checkout.
Route LLM calls across 9 providers with parallel racing, circuit breakers, and cost tracking — never let one dead provider kill your agent. Sends each prompt to every healthy provider at once (Groq, Cerebras, Mistral, OpenRouter, NVIDIA, Cloudflare, Portkey, SambaNova, LiteLLM) and returns the fastest answer that passes validation. Weighted load-balancing learns from live latency/success metrics in SQLite; circuit breakers sideline a failing provider for 5 minutes; prompt compression trims tokens before sending; responses are cached for 60 minutes; and every call's cost is logged with a session alert over $0.50. Keys load from `.env` or environment — no key is ever written to disk by the skill. Ships: `SKILL.md`, `multi-provider-failover-v2.py`, `metrics.py`, `validator.py`, `compressor.py`, `cost_tracker.py`, `model-registry.json`. Python 3.10+, standard library only, `curl` on PATH. Honest limits: model names and prices in the registry are from May 2026 and may be stale — verify with each provider; the LiteLLM path needs a `litellm` venv; the encrypted-vault key path is optional and its vault script is not included.
manteclaw
0
Sales
$0
Revenue
—
Rating
1
Products
Loading...