Skip to main content

Configuration Reference

The ICE engine is configured exclusively via environment variables. These parameters govern memory allocation, scaling, security, and document processing.

1. Core Infrastructure

VariableDescriptionDefault
ICE_PORTPort the engine listens on.8000
ICE_WORKERSNumber of Uvicorn worker processes.4
DATABASE_URLPostgreSQL connection string (requires pgvector).N/A
REDIS_URLRedis connection string for Hot-Cache.N/A
ICE_LICENSE_JWTLicense key for Pro/Enterprise tiers.N/A
ICE_UPLOAD_DIRSandboxed directory for multimodal file ingestion./tmp/ice/uploads

2. Resource & Scaling

VariableDescriptionDefault
ICE_MEMORY_CAP_GBHard RAM limit for the ICE process.8
ICE_MAX_STITCH_CONCURRENCYMaximum concurrent context operations.16
ICE_RETENTION_DAYSData retention policy (days).30
ICE_PRE_PURGE_WEBHOOK_URLWebhook URL for cold storage archiving before deletion.N/A

3. Context & Prompt Engineering

VariableDescriptionDefault
ICE_MODEL_CONTEXT_WINDOWTotal virtual context budget (input + output).128000
ICE_POST_COMPRESSION_LIMITMaximum final prompt size sent to the LLM.100000
ICE_DEFAULT_OUTPUT_TOKENSDefault output token budget if not specified.4096
ICE_MAX_CONTINUATIONSMaximum recursive turns for Recursive Completion.5
ICE_DYNAMIC_OUTPUT_BUDGETEnables intent-based token budgeting.true

4. Multimodal Ingestion

VariableDescriptionDefault
ICE_CHUNK_SIZESize of text chunks for semantic partitioning.1024
ICE_CHUNK_OVERLAPOverlap between consecutive chunks.128
ICE_INGEST_PARSERDocument parser selection (docling, mineru).docling
ICE_INGEST_ENABLE_IMAGESProcess and embed images in documents.false
ICE_INGEST_ENABLE_TABLESConvert tables to Markdown during ingestion.true
ICE_INGEST_ENABLE_EQUATIONSExtract and encode mathematical equations.true
ICE_INGEST_ENABLE_COMPLIANCEEnable PII redaction and compliance scrubbing.false
ICE_INGEST_COMPLIANCE_LEVELCompliance depth (basic, strict, full).basic

5. LLM Routing

VariableDescriptionDefault
UPSTREAM_API_URLGlobal fallback URL for LLM requests.N/A
OPENAI_API_URLDedicated routing for OpenAI-compatible models.https://api.openai.com/v1
ANTHROPIC_API_URLDedicated routing for Anthropic models.https://api.anthropic.com/v1