Files
heretek-openclaw-core/CONFIGURATION_TEMPLATE.md
T
John Doe ac0f4ec3b7 docs: Add OpenClaw configuration template with 22 agent models
- openclaw.json: Working configuration from ~/.openclaw
- openclaw.json.example: Template with environment variable placeholders
- CONFIGURATION_TEMPLATE.md: Setup and troubleshooting guide
- Includes all 22 agent passthrough endpoints
- LiteLLM provider with claude-opus-4-6 primary model

Configuration includes:
- 22 autonomous agent models (agent/steward through agent/scribe)
- Claude Opus 4.6 as primary model
- Ollama local models for embedding and vision
- Proper LiteLLM gateway integration
- Environment variable placeholders for secure deployment
2026-04-02 07:07:23 -04:00

4.6 KiB

OpenClaw Configuration Template

This directory contains the OpenClaw configuration template files that can be used as a starting point for new deployments.

Files

File Description
openclaw.json Current working configuration (from ~/.openclaw/openclaw.json)
openclaw.json.example Template with placeholder values for new deployments
openclaw.json.template Backup of the working configuration

Quick Start

  1. Copy the example configuration:

    cp openclaw.json.example ~/.openclaw/openclaw.json
    
  2. Set environment variables:

    export GATEWAY_TOKEN="your-gateway-token"
    export DATABASE_URL="postgres://user:pass@host:5432/dbname"
    export EMBEDDING_MODEL="qwen3-embedding:8b"
    export LITELLM_MASTER_KEY="sk-your-litellm-master-key"
    
  3. Replace placeholders:

    sed -i "s|\${GATEWAY_TOKEN}|$GATEWAY_TOKEN|g" ~/.openclaw/openclaw.json
    sed -i "s|\${DATABASE_URL}|$DATABASE_URL|g" ~/.openclaw/openclaw.json
    sed -i "s|\${EMBEDDING_MODEL:-qwen3-embedding:8b}|$EMBEDDING_MODEL|g" ~/.openclaw/openclaw.json
    
  4. Validate configuration:

    openclaw doctor
    

Configuration Sections

Models

The template includes pre-configured models for:

  • LiteLLM Provider (24 models)

    • 22 Agent passthrough endpoints (agent/steward, agent/alpha, etc.)
    • 1 Primary model (claude-opus-4-6)
    • 1 Fallback model
  • Ollama Provider (11 models)

    • Various cloud and local models
    • Embedding models for RAG

Agents

All 22 Heretek agents are pre-configured:

Agent Role Model
steward Orchestrator litellm/agent/steward
alpha Triad member litellm/agent/alpha
beta Triad member litellm/agent/beta
charlie Triad member litellm/agent/charlie
examiner Interrogator litellm/agent/examiner
explorer Scout litellm/agent/explorer
sentinel Guardian litellm/agent/sentinel
sentinel-prime Prime Guardian litellm/agent/sentinel-prime
coder Artisan litellm/agent/coder
dreamer Visionary litellm/agent/dreamer
empath Diplomat litellm/agent/empath
historian Archivist litellm/agent/historian
arbiter Judge litellm/agent/arbiter
catalyst Innovator litellm/agent/catalyst
chronos Timekeeper litellm/agent/chronos
coordinator Manager litellm/agent/coordinator
echo Reflector litellm/agent/echo
habit-forge Builder litellm/agent/habit-forge
metis Strategist litellm/agent/metis
nexus Connector litellm/agent/nexus
perceiver Sensor litellm/agent/perceiver
prism Analyzer litellm/agent/prism

Gateway

Gateway configuration includes:

  • Local mode binding
  • Token-based authentication
  • LAN binding with Tailscale support
  • Command restrictions for security

Plugins

Pre-configured plugins:

  • episodic-claw - Memory persistence
  • hybrid-search - Vector + keyword + graph search
  • consciousness - GWT/IIT/AST metrics
  • skill-extensions - Extended skill capabilities
  • liberation - Drive system management

Environment Variables

Variable Description Default
GATEWAY_TOKEN Gateway authentication token Required
DATABASE_URL PostgreSQL connection string Required
EMBEDDING_MODEL Embedding model for RAG qwen3-embedding:8b
LITELLM_MASTER_KEY LiteLLM API master key Required
OPENCLAW_VERSION OpenClaw version 2026.3.31

Validation

After configuring, validate with:

# Check configuration syntax
cat ~/.openclaw/openclaw.json | python3 -m json.tool > /dev/null && echo "Valid JSON"

# Run OpenClaw doctor
openclaw doctor

# Check LiteLLM connectivity
curl -H "Authorization: Bearer $LITELLM_MASTER_KEY" http://localhost:4000/health

Troubleshooting

Model Not Found Errors

If you see model_not_found errors:

  1. Ensure LiteLLM is running: docker ps | grep litellm
  2. Check models are loaded: curl -H "Authorization: Bearer $LITELLM_MASTER_KEY" http://localhost:4000/v1/models
  3. Verify model names match in both OpenClaw and LiteLLM configs

Authentication Errors

If you see authentication errors:

  1. Verify gateway token matches: grep token ~/.openclaw/openclaw.json
  2. Check LiteLLM master key: docker exec heretek-litellm env | grep LITELLM_MASTER_KEY
  3. Ensure API key format is correct (some require sk- prefix)

Updates

To update the template from a working configuration:

cp ~/.openclaw/openclaw.json /root/heretek/heretek-openclaw-core/openclaw.json

Then regenerate the example with placeholders.