Runtime Configuration

The Lucid base runtime (@lucid-fdn/agent-runtime) is a pre-built Docker image configured entirely through environment variables. It runs a minimal Express server on port 3100 that handles inference and automatic receipt creation.

Core Environment Variables

These are injected automatically when you launch through Lucid:

Variable	Required	Description
`LUCID_API_URL`	Yes	Lucid Layer API base URL
`LUCID_PASSPORT_ID`	Yes	Agent’s passport identity
`LUCID_API_KEY`	Yes	Authentication key for the Lucid API
`LUCID_WALLET_ADDRESS`	Auto	Agent wallet address (auto-created on launch)

Inference Configuration

Variable	Default	Description
`PROVIDER_URL`	—	OpenAI-compatible inference endpoint (TrustGate, Ollama, vLLM, LiteLLM)
`PROVIDER_API_KEY`	—	API key for the inference provider
`LUCID_MODEL`	—	Model to use (e.g., `openai/gpt-4o`)
`LUCID_PROMPT`	—	System prompt for the agent
`LUCID_TOOLS`	—	Comma-separated list of enabled tools
`TRUSTGATE_URL`	—	TrustGate URL (if using Lucid-managed inference)

Memory Configuration

Variable	Default	Description
`MEMORY_ENABLED`	`false`	Enable agent memory
`MEMORY_STORE`	`sqlite`	Storage backend: `sqlite`, `postgres`, or `memory`
`MEMORY_DB_PATH`	`./data/agents/{passport}/memory.db`	SQLite database path
`MEMORY_EXTRACTION_ENABLED`	`false`	Enable LLM-based memory extraction
`MEMORY_EMBEDDING_ENABLED`	`false`	Enable vector embeddings for semantic search
`MEMORY_EMBEDDING_PROVIDER`	`none`	Embedding provider: `openai`, `mock`, or `none`
`MEMORY_RECEIPTS_ENABLED`	`false`	Link memory writes to receipt chain
`MEMORY_PROJECTION_ENABLED`	`false`	Enable outbox-based projection

Deployment Control Plane

Variable	Default	Description
`DEPLOYMENT_STORE`	`postgres`	Deployment state store: `postgres` or `memory`
`DEPLOYMENT_CONTROL_PLANE`	`false`	Enable deployment control plane
`RECONCILER_POLL_MS`	`60000`	Reconciler polling interval
`RECONCILER_MAX_RETRIES`	`3`	Maximum retry attempts for failed deployments

DePIN and Anchoring

Variable	Default	Description
`DEPIN_UPLOAD_ENABLED`	`false`	Enable DePIN storage uploads
`DEPIN_PERMANENT_PROVIDER`	`mock`	Provider for permanent storage: `arweave`, `lighthouse`, `mock`
`DEPIN_EVOLVING_PROVIDER`	`mock`	Provider for evolving storage: `lighthouse`, `mock`
`ANCHOR_REGISTRY_STORE`	`memory`	Anchor registry: `postgres` or `memory`

NFT and Token Configuration

Variable	Default	Description
`NFT_PROVIDER`	`mock`	NFT minting backend: `token2022`, `metaplex-core`, `evm-erc721`, `mock`
`NFT_CHAINS`	—	Multi-chain minting targets: `solana-devnet,base`
`NFT_MINT_ON_CREATE`	`true`	Auto-mint NFT on passport creation
`TOKEN_LAUNCHER`	`mock`	Share token launcher: `direct-mint`, `genesis`, `mock`

Example: Full Configuration

# Core identity
LUCID_API_URL=https://api.lucid.foundation
LUCID_PASSPORT_ID=abc123
LUCID_API_KEY=lk_your_key

# Inference
PROVIDER_URL=https://trustgate.lucid.foundation
LUCID_MODEL=openai/gpt-4o
LUCID_PROMPT="You are a helpful trading assistant"

# Memory
MEMORY_ENABLED=true
MEMORY_STORE=sqlite

# Anchoring
DEPIN_UPLOAD_ENABLED=true
DEPIN_PERMANENT_PROVIDER=arweave

Launch an Agent

Configure an Agent

Deploy a Model

Compute Providers

Runtime Configuration

Core Environment Variables

Inference Configuration

Memory Configuration

Deployment Control Plane

DePIN and Anchoring

NFT and Token Configuration

Example: Full Configuration

​Core Environment Variables

​Inference Configuration

​Memory Configuration

​Deployment Control Plane

​DePIN and Anchoring

​NFT and Token Configuration

​Example: Full Configuration

Core Environment Variables

Inference Configuration

Memory Configuration

Deployment Control Plane

DePIN and Anchoring

NFT and Token Configuration

Example: Full Configuration