profbernardoj/NanoClaw/.claude/skills/add-morpheus

Add [REDACTED] — Decentralized Inference for NanoClaw

What This Does

Connects NanoClaw to the EverClaw proxy for decentralized AI inference via the [REDACTED] network. The proxy runs on the host machine and is reachable from Docker containers.

When To Use

Use /add-morpheus to:

• Enable hybrid mode (Claude orchestration + [REDACTED] inference)
• Check proxy health
• Switch default inference model

Setup

The EverClaw proxy must be running on the host (installed via setup.sh).

Environment Variables

Add to the NanoClaw container environment:

MORPHEUS_API_BASE=http://host.docker.internal:8083/v1
MORPHEUS_API_KEY=morpheus-local
MORPHEUS_DEFAULT_MODEL=glm-5

On Linux (native Docker), replace host.docker.internal with 172.17.0.1.

Docker Compose

If NanoClaw uses Docker Compose, add extra_hosts to ensure the alias resolves:

services:
  nanoclaw:
    extra_hosts:
      - "host.docker.internal:host-[REDACTED]"
    environment:
      - MORPHEUS_API_BASE=http://host.docker.internal:8083/v1
      - MORPHEUS_API_KEY=morpheus-local

Available Models

| Model | Use Case | |-------|----------| | glm-5 | Complex reasoning, coding, analysis (Opus 4.5-level) | | glm-4.7-flash | Fast responses, lightweight tasks | | kimi-k2.5 | General purpose | | qwen3-235b | Large context, multilingual |

Health Check

From inside the container:

curl -sf http://host.docker.internal:8083/health

From the host:

curl -sf http://127.0.0.1:8083/health

Hybrid Routing Strategy

• Use Claude for: orchestration, swarm coordination, complex multi-step tool use
• Use [REDACTED] for: text generation, summarization, research, sub-agent tasks
• Model selection is by name — pass glm-5 or glm-4.7-flash as the model parameter

Staking for Unlimited Inference

cd ~/.everclaw
node scripts/everclaw-wallet.mjs setup
node scripts/everclaw-wallet.mjs stake

MOR tokens are staked, not spent — returned when sessions close. Stake once, use forever.