Skip to main content
Glama
README.md
# Humane MCP
<!-- mcp-name: io.github.sammyboi81/humane -->

### *Waste less intelligence.*

Two capabilities that cut the real waste in AI — re-deriving what it already knew, and grinding one giant model serially through work a fleet of small ones could split — in **one install**, for **any model**.

- **Memory** (tamper-evident, local): `recall` before you re-derive, `remember` what you learn, `verify` the chain. Stop re-reading and re-figuring.
- **Orchestration** (provider-agnostic): `orchestrate` / `plan` / `dispatch` — right-sized parallel workers instead of a sledgehammer for every nail.

It **takes effect on connect**: the server hands your assistant instructions to recall-before-deriving and dispatch-instead-of-grinding. Bring your own key (Anthropic / OpenAI-compatible / Gemini). Free & Apache-2.0; paid upgrades add hosted/shared team memory + the governed team platform (`humane_upgrade`).

Twin of [ArkHive](https://arkhive.dondatabrain.com). Part of the #HumaneIntelligence movement.

## Install
```
pip install humane-mcp
# set one: ANTHROPIC_API_KEY  |  OPENAI_API_KEY  |  GEMINI_API_KEY
```

Tagline: **Waste less intelligence.**