fleet-memory/hindsight-docs/docs/developer/configuration.md

Configuration

Complete reference for configuring Hindsight services through environment variables.

Hindsight has two services, each with its own configuration prefix:

Service	Prefix	Description
API Service	HINDSIGHT_API_*	Core memory engine
Control Plane	HINDSIGHT_CP_*	Web UI

---

API Service

The API service handles all memory operations (retain, recall, reflect).

Database

Variable	Description	Default
HINDSIGHT_API_DATABASE_URL	PostgreSQL connection string	pg0 (embedded)
HINDSIGHT_API_DATABASE_SCHEMA	PostgreSQL schema name for tables	public
HINDSIGHT_API_RUN_MIGRATIONS_ON_STARTUP	Run database migrations on API startup	true

If not provided, the server uses embedded pg0 — convenient for development but not recommended for production.

The DATABASE_SCHEMA setting allows you to use a custom PostgreSQL schema instead of the default public schema. This is useful for:

• Multi-database setups where you want Hindsight tables in a dedicated schema
• Hosting platforms (e.g., Supabase) where public schema is reserved or shared
• Organizational preferences for schema naming conventions

# Example: Using a custom schema
export HINDSIGHT_API_DATABASE_URL=postgresql://user:pass@host:5432/dbname
export HINDSIGHT_API_DATABASE_SCHEMA=hindsight

Migrations will automatically create the schema if it doesn't exist and create all tables in the configured schema.

Database Connection Pool

Variable	Description	Default
HINDSIGHT_API_DB_POOL_MIN_SIZE	Minimum connections in the pool	5
HINDSIGHT_API_DB_POOL_MAX_SIZE	Maximum connections in the pool	100
HINDSIGHT_API_DB_COMMAND_TIMEOUT	PostgreSQL command timeout in seconds	60
HINDSIGHT_API_DB_ACQUIRE_TIMEOUT	Connection acquisition timeout in seconds	30

For high-concurrency workloads, increase DB_POOL_MAX_SIZE. Each concurrent recall/think operation can use 2-4 connections.

To run migrations manually (e.g., before starting the API), use the admin CLI:

hindsight-admin run-db-migration
# Or for a specific schema:
hindsight-admin run-db-migration --schema tenant_acme

Vector Extension

Variable	Description	Default
HINDSIGHT_API_VECTOR_EXTENSION	Vector extension to use: auto, pgvector, or vchord	auto

Hindsight supports two PostgreSQL vector extensions:

• pgvector: Standard extension, works well for most embeddings (up to ~2000 dimensions)
• vchord: Optimized for high-dimensional embeddings (3000+ dimensions), includes BM25 search

When set to auto (default), Hindsight automatically detects which extension is installed, preferring vchord if both are available.

When to use vchord:

• Using high-dimensional embeddings (e.g., text-embedding-3-large with 3072 dimensions)
• Need better performance with large embedding dimensions
• Want to use vchord's BM25 search capabilities

When to use pgvector:

• Using standard embedding dimensions (384-1536)
• Prefer the widely-adopted pgvector extension
• Simpler deployment (pgvector is more commonly available)

Switching extensions:

If you need to switch from one extension to another:

1. Set HINDSIGHT_API_VECTOR_EXTENSION to your desired extension (pgvector or vchord)
2. If your database has existing data, you'll get an error with migration instructions
3. For empty databases, indexes will be automatically recreated on startup

Text Search Extension

Variable	Description	Default
HINDSIGHT_API_TEXT_SEARCH_EXTENSION	Text search backend: native, vchord, or pg_textsearch	native

Hindsight supports three text search backends for BM25 keyword retrieval:

• native: PostgreSQL's built-in full-text search (tsvector + GIN indexes)
• vchord: VectorChord BM25 (bm25vector + BM25 indexes) - requires vchord_bm25 extension
• pg_textsearch: Timescale BM25 (text columns + BM25 indexes) - requires pg_textsearch extension

When to use native:

• Standard PostgreSQL deployment (no extra extensions)
• Simpler setup and wider compatibility
• Works well for most use cases

When to use vchord:

• Already using vchord for vector search (good integration)
• Want better BM25 ranking performance
• Need advanced tokenization (uses llmlingua2 tokenizer)

When to use pg_textsearch:

• Want industry-standard BM25 ranking with better relevance than native PostgreSQL
• Need efficient top-K queries with Block-Max WAND optimization
• Prefer lower memory footprint compared to vchord
• Already using Timescale or have pg_textsearch available

Switching backends:

To switch between backends:

1. Set HINDSIGHT_API_TEXT_SEARCH_EXTENSION to your desired backend (native, vchord, or pg_textsearch)
2. If your database has existing data, you'll get an error with migration instructions
3. For empty databases, the columns/indexes will be automatically recreated on startup

Note: VectorChord uses the llmlingua2 tokenizer for multilingual support, while native and pg_textsearch use PostgreSQL's English tokenizer.

LLM Provider

Variable	Description	Default
HINDSIGHT_API_LLM_PROVIDER	Provider: openai, openai-codex, claude-code, anthropic, gemini, groq, ollama, lmstudio, vertexai	openai
HINDSIGHT_API_LLM_API_KEY	API key for LLM provider	-
HINDSIGHT_API_LLM_MODEL	Model name	gpt-5-mini
HINDSIGHT_API_LLM_BASE_URL	Custom LLM endpoint	Provider default
HINDSIGHT_API_LLM_MAX_CONCURRENT	Max concurrent LLM requests	32
HINDSIGHT_API_LLM_MAX_RETRIES	Max retry attempts for LLM API calls	10
HINDSIGHT_API_LLM_INITIAL_BACKOFF	Initial retry backoff in seconds (exponential backoff)	1.0
HINDSIGHT_API_LLM_MAX_BACKOFF	Max retry backoff cap in seconds	60.0
HINDSIGHT_API_LLM_TIMEOUT	LLM request timeout in seconds	120
HINDSIGHT_API_LLM_GROQ_SERVICE_TIER	Groq service tier: on_demand, flex, auto	auto

Provider Examples

# Groq (recommended for fast inference)
export HINDSIGHT_API_LLM_PROVIDER=groq
export HINDSIGHT_API_LLM_API_KEY=gsk_xxxxxxxxxxxx
export HINDSIGHT_API_LLM_MODEL=openai/gpt-oss-20b
# For free tier users: override to on_demand if you get service_tier errors
# export HINDSIGHT_API_LLM_GROQ_SERVICE_TIER=on_demand

# OpenAI
export HINDSIGHT_API_LLM_PROVIDER=openai
export HINDSIGHT_API_LLM_API_KEY=sk-xxxxxxxxxxxx
export HINDSIGHT_API_LLM_MODEL=gpt-4o

# Gemini
export HINDSIGHT_API_LLM_PROVIDER=gemini
export HINDSIGHT_API_LLM_API_KEY=xxxxxxxxxxxx
export HINDSIGHT_API_LLM_MODEL=gemini-2.0-flash

# Anthropic
export HINDSIGHT_API_LLM_PROVIDER=anthropic
export HINDSIGHT_API_LLM_API_KEY=sk-ant-xxxxxxxxxxxx
export HINDSIGHT_API_LLM_MODEL=claude-sonnet-4-20250514

# Vertex AI (Google Cloud - uses native genai SDK)
export HINDSIGHT_API_LLM_PROVIDER=vertexai
export HINDSIGHT_API_LLM_MODEL=gemini-2.0-flash-001
export HINDSIGHT_API_LLM_VERTEXAI_PROJECT_ID=your-gcp-project-id
export HINDSIGHT_API_LLM_VERTEXAI_REGION=us-central1
# Optional: use ADC (gcloud auth application-default login) or provide service account key:
# export HINDSIGHT_API_LLM_VERTEXAI_SERVICE_ACCOUNT_KEY=/path/to/service-account-key.json

# Ollama (local, no API key)
export HINDSIGHT_API_LLM_PROVIDER=ollama
export HINDSIGHT_API_LLM_BASE_URL=http://localhost:11434/v1
export HINDSIGHT_API_LLM_MODEL=llama3

# LM Studio (local, no API key)
export HINDSIGHT_API_LLM_PROVIDER=lmstudio
export HINDSIGHT_API_LLM_BASE_URL=http://localhost:1234/v1
export HINDSIGHT_API_LLM_MODEL=your-local-model

# OpenAI-compatible endpoint
export HINDSIGHT_API_LLM_PROVIDER=openai
export HINDSIGHT_API_LLM_BASE_URL=https://your-endpoint.com/v1
export HINDSIGHT_API_LLM_API_KEY=your-api-key
export HINDSIGHT_API_LLM_MODEL=your-model-name

# OpenAI Codex (ChatGPT Plus/Pro subscription - uses OAuth, no API key needed)
export HINDSIGHT_API_LLM_PROVIDER=openai-codex
export HINDSIGHT_API_LLM_MODEL=gpt-5.2-codex
# No API key needed - uses OAuth tokens from ~/.codex/auth.json

# Claude Code (Claude Pro/Max subscription - uses OAuth, no API key needed)
export HINDSIGHT_API_LLM_PROVIDER=claude-code
export HINDSIGHT_API_LLM_MODEL=claude-sonnet-4-5-20250929
# No API key needed - uses claude auth login credentials

:::tip OpenAI Codex & Claude Code Setup For detailed setup instructions for OpenAI Codex (ChatGPT Plus/Pro) and Claude Code (Claude Pro/Max), see the Models documentation. :::

Vertex AI Setup

Google Cloud's Vertex AI provides access to Gemini models via the native Google GenAI SDK. Hindsight supports two authentication methods:

Prerequisites:

• GCP project with Vertex AI API enabled
• IAM role roles/aiplatform.user for your credentials

Environment Variables:

Variable	Description	Required
HINDSIGHT_API_LLM_VERTEXAI_PROJECT_ID	Your GCP project ID	Yes
HINDSIGHT_API_LLM_VERTEXAI_REGION	GCP region (e.g., us-central1)	No (default: us-central1)
HINDSIGHT_API_LLM_VERTEXAI_SERVICE_ACCOUNT_KEY	Path to service account JSON key file	No (uses ADC if not set)

Authentication Methods:

1. Application Default Credentials (ADC) - Recommended for development

   # Setup ADC
   gcloud auth application-default login
   
   # Configure Hindsight
   export HINDSIGHT_API_LLM_PROVIDER=vertexai
   export HINDSIGHT_API_LLM_MODEL=gemini-2.0-flash-001
   export HINDSIGHT_API_LLM_VERTEXAI_PROJECT_ID=your-project-id
   

2. Service Account Key - Recommended for production

   # Create service account and download key
   gcloud iam service-accounts create hindsight-api
   gcloud projects add-iam-policy-binding your-project-id \
     --member="serviceAccount:hindsight-api@your-project-id.iam.gserviceaccount.com" \
     --role="roles/aiplatform.user"
   gcloud iam service-accounts keys create key.json \
     --iam-account=hindsight-api@your-project-id.iam.gserviceaccount.com
   
   # Configure Hindsight
   export HINDSIGHT_API_LLM_PROVIDER=vertexai
   export HINDSIGHT_API_LLM_MODEL=gemini-2.0-flash-001
   export HINDSIGHT_API_LLM_VERTEXAI_PROJECT_ID=your-project-id
   export HINDSIGHT_API_LLM_VERTEXAI_SERVICE_ACCOUNT_KEY=/path/to/key.json
   

Notes:

• Model names can optionally include the google/ prefix (e.g., google/gemini-2.0-flash-001) - it will be stripped automatically
• The native SDK handles token refresh automatically
• Uses service account credentials if provided, otherwise falls back to ADC

Per-Operation LLM Configuration

Different memory operations have different requirements. Retain (fact extraction) benefits from models with strong structured output capabilities, while Reflect (reasoning/response generation) can use lighter, faster models. Configure separate LLM models for each operation to optimize for cost and performance.

Variable	Description	Default
HINDSIGHT_API_RETAIN_LLM_PROVIDER	LLM provider for retain operations	Falls back to HINDSIGHT_API_LLM_PROVIDER
HINDSIGHT_API_RETAIN_LLM_API_KEY	API key for retain LLM	Falls back to HINDSIGHT_API_LLM_API_KEY
HINDSIGHT_API_RETAIN_LLM_MODEL	Model for retain operations	Falls back to HINDSIGHT_API_LLM_MODEL
HINDSIGHT_API_RETAIN_LLM_BASE_URL	Base URL for retain LLM	Falls back to HINDSIGHT_API_LLM_BASE_URL
HINDSIGHT_API_RETAIN_LLM_MAX_CONCURRENT	Max concurrent requests for retain	Falls back to HINDSIGHT_API_LLM_MAX_CONCURRENT
HINDSIGHT_API_RETAIN_LLM_MAX_RETRIES	Max retries for retain	Falls back to HINDSIGHT_API_LLM_MAX_RETRIES
HINDSIGHT_API_RETAIN_LLM_INITIAL_BACKOFF	Initial backoff for retain retries (seconds)	Falls back to HINDSIGHT_API_LLM_INITIAL_BACKOFF
HINDSIGHT_API_RETAIN_LLM_MAX_BACKOFF	Max backoff cap for retain retries (seconds)	Falls back to HINDSIGHT_API_LLM_MAX_BACKOFF
HINDSIGHT_API_RETAIN_LLM_TIMEOUT	Timeout for retain requests (seconds)	Falls back to HINDSIGHT_API_LLM_TIMEOUT
HINDSIGHT_API_REFLECT_LLM_PROVIDER	LLM provider for reflect operations	Falls back to HINDSIGHT_API_LLM_PROVIDER
HINDSIGHT_API_REFLECT_LLM_API_KEY	API key for reflect LLM	Falls back to HINDSIGHT_API_LLM_API_KEY
HINDSIGHT_API_REFLECT_LLM_MODEL	Model for reflect operations	Falls back to HINDSIGHT_API_LLM_MODEL
HINDSIGHT_API_REFLECT_LLM_BASE_URL	Base URL for reflect LLM	Falls back to HINDSIGHT_API_LLM_BASE_URL
HINDSIGHT_API_REFLECT_LLM_MAX_CONCURRENT	Max concurrent requests for reflect	Falls back to HINDSIGHT_API_LLM_MAX_CONCURRENT
HINDSIGHT_API_REFLECT_LLM_MAX_RETRIES	Max retries for reflect	Falls back to HINDSIGHT_API_LLM_MAX_RETRIES
HINDSIGHT_API_REFLECT_LLM_INITIAL_BACKOFF	Initial backoff for reflect retries (seconds)	Falls back to HINDSIGHT_API_LLM_INITIAL_BACKOFF
HINDSIGHT_API_REFLECT_LLM_MAX_BACKOFF	Max backoff cap for reflect retries (seconds)	Falls back to HINDSIGHT_API_LLM_MAX_BACKOFF
HINDSIGHT_API_REFLECT_LLM_TIMEOUT	Timeout for reflect requests (seconds)	Falls back to HINDSIGHT_API_LLM_TIMEOUT
HINDSIGHT_API_CONSOLIDATION_LLM_PROVIDER	LLM provider for observation consolidation	Falls back to HINDSIGHT_API_LLM_PROVIDER
HINDSIGHT_API_CONSOLIDATION_LLM_API_KEY	API key for consolidation LLM	Falls back to HINDSIGHT_API_LLM_API_KEY
HINDSIGHT_API_CONSOLIDATION_LLM_MODEL	Model for consolidation operations	Falls back to HINDSIGHT_API_LLM_MODEL
HINDSIGHT_API_CONSOLIDATION_LLM_BASE_URL	Base URL for consolidation LLM	Falls back to HINDSIGHT_API_LLM_BASE_URL
HINDSIGHT_API_CONSOLIDATION_LLM_MAX_CONCURRENT	Max concurrent requests for consolidation	Falls back to HINDSIGHT_API_LLM_MAX_CONCURRENT
HINDSIGHT_API_CONSOLIDATION_LLM_MAX_RETRIES	Max retries for consolidation	Falls back to HINDSIGHT_API_LLM_MAX_RETRIES
HINDSIGHT_API_CONSOLIDATION_LLM_INITIAL_BACKOFF	Initial backoff for consolidation retries (seconds)	Falls back to HINDSIGHT_API_LLM_INITIAL_BACKOFF
HINDSIGHT_API_CONSOLIDATION_LLM_MAX_BACKOFF	Max backoff cap for consolidation retries (seconds)	Falls back to HINDSIGHT_API_LLM_MAX_BACKOFF
HINDSIGHT_API_CONSOLIDATION_LLM_TIMEOUT	Timeout for consolidation requests (seconds)	Falls back to HINDSIGHT_API_LLM_TIMEOUT

:::tip When to Use Per-Operation Config

• Retain: Use models with strong structured output (e.g., GPT-4o, Claude) for accurate fact extraction
• Reflect: Use faster/cheaper models (e.g., GPT-4o-mini, Groq) for reasoning and response generation
• Recall: Does not use LLM (pure retrieval), so no configuration needed :::

Example: Separate Models for Retain and Reflect

# Default LLM (used as fallback)
export HINDSIGHT_API_LLM_PROVIDER=openai
export HINDSIGHT_API_LLM_API_KEY=sk-xxxxxxxxxxxx
export HINDSIGHT_API_LLM_MODEL=gpt-4o

# Use GPT-4o for retain (strong structured output)
export HINDSIGHT_API_RETAIN_LLM_MODEL=gpt-4o

# Use faster/cheaper model for reflect
export HINDSIGHT_API_REFLECT_LLM_PROVIDER=groq
export HINDSIGHT_API_REFLECT_LLM_API_KEY=gsk_xxxxxxxxxxxx
export HINDSIGHT_API_REFLECT_LLM_MODEL=llama-3.3-70b-versatile

Example: Tuning Retry Behavior for Rate-Limited APIs

# For Anthropic with tight rate limits (10k output tokens/minute)
export HINDSIGHT_API_LLM_PROVIDER=anthropic
export HINDSIGHT_API_LLM_API_KEY=sk-ant-xxxxxxxxxxxx
export HINDSIGHT_API_LLM_MODEL=claude-sonnet-4-20250514

# Reduce concurrent requests for retain to avoid rate limits
export HINDSIGHT_API_RETAIN_LLM_MAX_CONCURRENT=3

# Fail faster with fewer retries
export HINDSIGHT_API_RETAIN_LLM_MAX_RETRIES=3

# Or increase backoff times to wait out rate limit windows
export HINDSIGHT_API_RETAIN_LLM_INITIAL_BACKOFF=2.0  # Start at 2s instead of 1s
export HINDSIGHT_API_RETAIN_LLM_MAX_BACKOFF=120.0    # Cap at 2min instead of 1min

Embeddings

Variable	Description	Default
HINDSIGHT_API_EMBEDDINGS_PROVIDER	Provider: local, tei, openai, cohere, or litellm	local
HINDSIGHT_API_EMBEDDINGS_LOCAL_MODEL	Model for local provider	BAAI/bge-small-en-v1.5
HINDSIGHT_API_EMBEDDINGS_LOCAL_TRUST_REMOTE_CODE	Allow loading models with custom code (security risk, disabled by default)	false
HINDSIGHT_API_EMBEDDINGS_TEI_URL	TEI server URL	-
HINDSIGHT_API_EMBEDDINGS_OPENAI_API_KEY	OpenAI API key (falls back to HINDSIGHT_API_LLM_API_KEY)	-
HINDSIGHT_API_EMBEDDINGS_OPENAI_MODEL	OpenAI embedding model	text-embedding-3-small
HINDSIGHT_API_EMBEDDINGS_OPENAI_BASE_URL	Custom base URL for OpenAI-compatible API (e.g., Azure OpenAI)	-
HINDSIGHT_API_EMBEDDINGS_COHERE_API_KEY	Cohere API key for embeddings	-
HINDSIGHT_API_EMBEDDINGS_COHERE_MODEL	Cohere embedding model	embed-english-v3.0
HINDSIGHT_API_EMBEDDINGS_COHERE_BASE_URL	Custom base URL for Cohere-compatible API (e.g., Azure-hosted)	-
HINDSIGHT_API_EMBEDDINGS_LITELLM_API_BASE	LiteLLM proxy base URL for embeddings	http://localhost:4000
HINDSIGHT_API_EMBEDDINGS_LITELLM_API_KEY	LiteLLM proxy API key for embeddings (optional, depends on proxy config)	-
HINDSIGHT_API_EMBEDDINGS_LITELLM_MODEL	LiteLLM embedding model (use provider prefix, e.g., cohere/embed-english-v3.0)	text-embedding-3-small

# Local (default) - uses SentenceTransformers
export HINDSIGHT_API_EMBEDDINGS_PROVIDER=local
export HINDSIGHT_API_EMBEDDINGS_LOCAL_MODEL=BAAI/bge-small-en-v1.5

# Local with custom model requiring trust_remote_code
# WARNING: Only enable trust_remote_code for models you trust (security risk)
# export HINDSIGHT_API_EMBEDDINGS_LOCAL_MODEL=your-custom-model
# export HINDSIGHT_API_EMBEDDINGS_LOCAL_TRUST_REMOTE_CODE=true

# OpenAI - cloud-based embeddings
export HINDSIGHT_API_EMBEDDINGS_PROVIDER=openai
export HINDSIGHT_API_EMBEDDINGS_OPENAI_API_KEY=sk-xxxxxxxxxxxx  # or reuses HINDSIGHT_API_LLM_API_KEY
export HINDSIGHT_API_EMBEDDINGS_OPENAI_MODEL=text-embedding-3-small  # 1536 dimensions

# Azure OpenAI - embeddings via Azure endpoint
export HINDSIGHT_API_EMBEDDINGS_PROVIDER=openai
export HINDSIGHT_API_EMBEDDINGS_OPENAI_API_KEY=your-azure-api-key
export HINDSIGHT_API_EMBEDDINGS_OPENAI_MODEL=text-embedding-3-small
export HINDSIGHT_API_EMBEDDINGS_OPENAI_BASE_URL=https://your-resource.openai.azure.com/openai/deployments/your-deployment

# TEI - HuggingFace Text Embeddings Inference (recommended for production)
export HINDSIGHT_API_EMBEDDINGS_PROVIDER=tei
export HINDSIGHT_API_EMBEDDINGS_TEI_URL=http://localhost:8080

# Cohere - cloud-based embeddings
export HINDSIGHT_API_EMBEDDINGS_PROVIDER=cohere
export HINDSIGHT_API_EMBEDDINGS_COHERE_API_KEY=your-api-key
export HINDSIGHT_API_EMBEDDINGS_COHERE_MODEL=embed-english-v3.0  # 1024 dimensions

# Azure-hosted Cohere - embeddings via custom endpoint
export HINDSIGHT_API_EMBEDDINGS_PROVIDER=cohere
export HINDSIGHT_API_EMBEDDINGS_COHERE_API_KEY=your-azure-api-key
export HINDSIGHT_API_EMBEDDINGS_COHERE_MODEL=embed-english-v3.0
export HINDSIGHT_API_EMBEDDINGS_COHERE_BASE_URL=https://your-azure-cohere-endpoint.com

# LiteLLM proxy - unified gateway for multiple providers
export HINDSIGHT_API_EMBEDDINGS_PROVIDER=litellm
export HINDSIGHT_API_EMBEDDINGS_LITELLM_API_BASE=http://localhost:4000
export HINDSIGHT_API_EMBEDDINGS_LITELLM_API_KEY=your-litellm-key  # optional
export HINDSIGHT_API_EMBEDDINGS_LITELLM_MODEL=text-embedding-3-small  # or cohere/embed-english-v3.0

Embedding Dimensions

Hindsight automatically detects the embedding dimension from the model at startup and adjusts the database schema accordingly. The default model (BAAI/bge-small-en-v1.5) produces 384-dimensional vectors, while OpenAI models produce 1536 or 3072 dimensions.

:::warning Dimension Changes Once memories are stored, you cannot change the embedding dimension without losing data. If you need to switch to a model with different dimensions:

1. Empty database: The schema is adjusted automatically on startup
2. Existing data: Either delete all memories first, or use a model with matching dimensions

Supported OpenAI embedding dimensions:

• text-embedding-3-small: 1536 dimensions
• text-embedding-3-large: 3072 dimensions
• text-embedding-ada-002: 1536 dimensions (legacy) :::

Reranker

Variable	Description	Default
HINDSIGHT_API_RERANKER_PROVIDER	Provider: local, tei, cohere, flashrank, litellm, or rrf	local
HINDSIGHT_API_RERANKER_LOCAL_MODEL	Model for local provider	cross-encoder/ms-marco-MiniLM-L-6-v2
HINDSIGHT_API_RERANKER_LOCAL_MAX_CONCURRENT	Max concurrent local reranking (prevents CPU thrashing under load)	4
HINDSIGHT_API_RERANKER_LOCAL_TRUST_REMOTE_CODE	Allow loading models with custom code (security risk, disabled by default)	false
HINDSIGHT_API_RERANKER_TEI_URL	TEI server URL	-
HINDSIGHT_API_RERANKER_TEI_BATCH_SIZE	Batch size for TEI reranking	128
HINDSIGHT_API_RERANKER_TEI_MAX_CONCURRENT	Max concurrent TEI reranking requests	8
HINDSIGHT_API_RERANKER_COHERE_API_KEY	Cohere API key for reranking	-
HINDSIGHT_API_RERANKER_COHERE_MODEL	Cohere rerank model	rerank-english-v3.0
HINDSIGHT_API_RERANKER_COHERE_BASE_URL	Custom base URL for Cohere-compatible API (e.g., Azure-hosted)	-
HINDSIGHT_API_RERANKER_LITELLM_API_BASE	LiteLLM proxy base URL for reranking	http://localhost:4000
HINDSIGHT_API_RERANKER_LITELLM_API_KEY	LiteLLM proxy API key for reranking (optional, depends on proxy config)	-
HINDSIGHT_API_RERANKER_LITELLM_MODEL	LiteLLM rerank model (use provider prefix, e.g., cohere/rerank-english-v3.0)	cohere/rerank-english-v3.0
HINDSIGHT_API_RERANKER_FLASHRANK_MODEL	FlashRank model for fast CPU-based reranking	ms-marco-MiniLM-L-12-v2
HINDSIGHT_API_RERANKER_FLASHRANK_CACHE_DIR	Cache directory for FlashRank models	System default

# Local (default) - uses SentenceTransformers CrossEncoder
export HINDSIGHT_API_RERANKER_PROVIDER=local
export HINDSIGHT_API_RERANKER_LOCAL_MODEL=cross-encoder/ms-marco-MiniLM-L-6-v2

# Local with custom model requiring trust_remote_code (e.g., jina-reranker-v2)
# WARNING: Only enable trust_remote_code for models you trust (security risk)
export HINDSIGHT_API_RERANKER_PROVIDER=local
export HINDSIGHT_API_RERANKER_LOCAL_MODEL=jinaai/jina-reranker-v2-base-multilingual
export HINDSIGHT_API_RERANKER_LOCAL_TRUST_REMOTE_CODE=true

# TEI - for high-performance inference
export HINDSIGHT_API_RERANKER_PROVIDER=tei
export HINDSIGHT_API_RERANKER_TEI_URL=http://localhost:8081

# Cohere - cloud-based reranking
export HINDSIGHT_API_RERANKER_PROVIDER=cohere
export HINDSIGHT_API_RERANKER_COHERE_API_KEY=your-api-key
export HINDSIGHT_API_RERANKER_COHERE_MODEL=rerank-english-v3.0

# Azure-hosted Cohere - reranking via custom endpoint
export HINDSIGHT_API_RERANKER_PROVIDER=cohere
export HINDSIGHT_API_RERANKER_COHERE_API_KEY=your-azure-api-key
export HINDSIGHT_API_RERANKER_COHERE_MODEL=rerank-english-v3.0
export HINDSIGHT_API_RERANKER_COHERE_BASE_URL=https://your-azure-cohere-endpoint.com

# LiteLLM proxy - unified gateway for multiple reranking providers
export HINDSIGHT_API_RERANKER_PROVIDER=litellm
export HINDSIGHT_API_RERANKER_LITELLM_API_BASE=http://localhost:4000
export HINDSIGHT_API_RERANKER_LITELLM_API_KEY=your-litellm-key  # optional
export HINDSIGHT_API_RERANKER_LITELLM_MODEL=cohere/rerank-english-v3.0  # or voyage/rerank-2, together_ai/...

LiteLLM supports multiple reranking providers via the /rerank endpoint:

• Cohere (cohere/rerank-english-v3.0, cohere/rerank-multilingual-v3.0)
• Together AI (together_ai/...)
• Voyage AI (voyage/rerank-2)
• Jina AI (jina_ai/...)
• AWS Bedrock (bedrock/...)

Authentication

By default, Hindsight runs without authentication. For production deployments, enable API key authentication using the built-in tenant extension:

# Enable the built-in API key authentication
export HINDSIGHT_API_TENANT_EXTENSION=hindsight_api.extensions.builtin.tenant:ApiKeyTenantExtension
export HINDSIGHT_API_TENANT_API_KEY=your-secret-api-key

When enabled, all requests must include the API key in the Authorization header:

curl -H "Authorization: Bearer your-secret-api-key" \
  http://localhost:8888/v1/default/banks

Requests without a valid API key receive a 401 Unauthorized response.

:::tip Custom Authentication For advanced authentication (JWT, OAuth, multi-tenant schemas), implement a custom TenantExtension. See the Extensions documentation for details. :::

Server

Variable	Description	Default
HINDSIGHT_API_HOST	Bind address	0.0.0.0
HINDSIGHT_API_PORT	Server port	8888
HINDSIGHT_API_BASE_PATH	Base path for API when behind reverse proxy (e.g., /hindsight)	"" (root)
HINDSIGHT_API_WORKERS	Number of uvicorn worker processes	1
HINDSIGHT_API_LOG_LEVEL	Log level: debug, info, warning, error	info
HINDSIGHT_API_LOG_FORMAT	Log format: text or json (structured logging for cloud platforms)	text
HINDSIGHT_API_MCP_ENABLED	Enable MCP server at /mcp/{bank_id}/	true

Retrieval

Variable	Description	Default
HINDSIGHT_API_GRAPH_RETRIEVER	Graph retrieval algorithm: link_expansion, mpfp, or bfs	link_expansion
HINDSIGHT_API_RECALL_MAX_CONCURRENT	Max concurrent recall operations per worker (backpressure)	32
HINDSIGHT_API_RECALL_CONNECTION_BUDGET	Max concurrent DB connections per recall operation	4
HINDSIGHT_API_RERANKER_MAX_CANDIDATES	Max candidates to rerank per recall (RRF pre-filters the rest)	300
HINDSIGHT_API_MPFP_TOP_K_NEIGHBORS	Fan-out limit per node in MPFP graph traversal	20
HINDSIGHT_API_MENTAL_MODEL_REFRESH_CONCURRENCY	Max concurrent mental model refreshes	8

Graph Retrieval Algorithms

• link_expansion (default): Fast, simple graph expansion from semantic seeds via entity co-occurrence and causal links. Target latency under 100ms. Recommended for most use cases.
• mpfp: Multi-Path Fact Propagation - iterative graph traversal with activation spreading. More thorough but slower.
• bfs: Breadth-first search from seed facts. Simple but less effective for large graphs.

Retain

Controls the retain (memory ingestion) pipeline.

Variable	Description	Default
HINDSIGHT_API_RETAIN_MAX_COMPLETION_TOKENS	Max completion tokens for fact extraction LLM calls	64000
HINDSIGHT_API_RETAIN_CHUNK_SIZE	Max characters per chunk for fact extraction. Larger chunks extract fewer LLM calls but may lose context.	3000
HINDSIGHT_API_RETAIN_EXTRACTION_MODE	Fact extraction mode: concise, verbose, or custom	concise
HINDSIGHT_API_RETAIN_CUSTOM_INSTRUCTIONS	Custom extraction guidelines (only used when mode is custom)	-
HINDSIGHT_API_RETAIN_EXTRACT_CAUSAL_LINKS	Extract causal relationships between facts	true

Extraction Modes

The extraction mode controls how aggressively facts are extracted from content:

• concise (default): Selective extraction that focuses on significant, long-term valuable facts. Filters out greetings, filler, and trivial information. Produces fewer but higher-quality facts with better performance.

• verbose: Detailed extraction that captures every piece of information with maximum verbosity. Produces more facts with extensive detail but slower performance and higher token usage.

• custom: Inject your own extraction guidelines while keeping the structural parts of the prompt (output format, coreference resolution, temporal handling, etc.) intact. Useful for A/B testing different extraction strategies or domain-specific customization.

Example: Custom Extraction Mode

# Set mode to custom
export HINDSIGHT_API_RETAIN_EXTRACTION_MODE=custom

# Define custom guidelines (multi-line is fine)
export HINDSIGHT_API_RETAIN_CUSTOM_INSTRUCTIONS="ONLY extract facts that are:
✅ Technical decisions and their rationale
✅ Architecture patterns and design choices
✅ Performance metrics and benchmarks
✅ Code reviews and feedback

DO NOT extract:
❌ Generic greetings or pleasantries
❌ Process chatter (\"let me check\", \"one moment\")
❌ Repeated information already captured

CONSOLIDATE related technical discussions into ONE fact when possible.

Ask yourself: 'Would this technical context be useful in 6 months?' If no, skip it."

Observations (Experimental)

Observations are consolidated knowledge synthesized from facts.

Variable	Description	Default
HINDSIGHT_API_ENABLE_OBSERVATIONS	Enable observation consolidation	true
HINDSIGHT_API_CONSOLIDATION_BATCH_SIZE	Memories to load per batch (internal optimization)	50
HINDSIGHT_API_CONSOLIDATION_MAX_TOKENS	Max tokens for recall when finding related observations during consolidation	1024

Reflect

Variable	Description	Default
HINDSIGHT_API_REFLECT_MAX_ITERATIONS	Max tool call iterations before forcing a response	10

MCP Server

Configuration for MCP server endpoints.

Variable	Description	Default
HINDSIGHT_API_MCP_ENABLED	Enable MCP server at /mcp/{bank_id}/	true
HINDSIGHT_API_MCP_AUTH_TOKEN	Bearer token for MCP authentication (optional)	-
HINDSIGHT_API_MCP_LOCAL_BANK_ID	Memory bank ID for local MCP	mcp
HINDSIGHT_API_MCP_INSTRUCTIONS	Additional instructions appended to retain/recall tool descriptions	-

MCP Authentication:

By default, the MCP endpoint is open. For production deployments, set HINDSIGHT_API_MCP_AUTH_TOKEN to require Bearer token authentication:

export HINDSIGHT_API_MCP_AUTH_TOKEN=your-secret-token

Clients must then include the token in the Authorization header. See MCP Server documentation for details.

Local MCP instructions:

# Example: instruct MCP to also store assistant actions
export HINDSIGHT_API_MCP_INSTRUCTIONS="Also store every action you take, including tool calls and decisions made."

Distributed Workers

Configuration for background task processing. By default, the API processes tasks internally. For high-throughput deployments, run dedicated workers. See Services - Worker Service for details.

Variable	Description	Default
HINDSIGHT_API_WORKER_ENABLED	Enable internal worker in API process	true
HINDSIGHT_API_WORKER_ID	Unique worker identifier	hostname
HINDSIGHT_API_WORKER_POLL_INTERVAL_MS	Database polling interval in milliseconds	500
HINDSIGHT_API_WORKER_MAX_RETRIES	Max retries before marking task failed	3
HINDSIGHT_API_WORKER_HTTP_PORT	HTTP port for worker metrics/health (worker CLI only)	8889
HINDSIGHT_API_WORKER_MAX_SLOTS	Maximum concurrent tasks per worker	10
HINDSIGHT_API_WORKER_CONSOLIDATION_MAX_SLOTS	Maximum concurrent consolidation tasks per worker	2

Performance Optimization

Variable	Description	Default
HINDSIGHT_API_SKIP_LLM_VERIFICATION	Skip LLM connection check on startup	false
HINDSIGHT_API_LAZY_RERANKER	Lazy-load reranker model (faster startup)	false

Programmatic Configuration

You can also configure the API programmatically using MemoryEngine.from_env():

from hindsight_api import MemoryEngine

memory = MemoryEngine.from_env()
await memory.initialize()

---

Observability & Tracing

Hindsight provides OpenTelemetry-based observability for LLM calls, conforming to GenAI semantic conventions.

OpenTelemetry Tracing

Variable	Description	Default
HINDSIGHT_API_OTEL_TRACES_ENABLED	Enable distributed tracing for LLM calls	false
HINDSIGHT_API_OTEL_EXPORTER_OTLP_ENDPOINT	OTLP endpoint URL (e.g., Grafana LGTM, Langfuse, etc.)	-
HINDSIGHT_API_OTEL_EXPORTER_OTLP_HEADERS	Headers for OTLP exporter (format: "key1=value1,key2=value2")	-
HINDSIGHT_API_OTEL_SERVICE_NAME	Service name for traces	hindsight-api
HINDSIGHT_API_OTEL_DEPLOYMENT_ENVIRONMENT	Deployment environment name (e.g., development, staging, production)	development

Features:

• Full prompts and completions recorded as events
• Token usage tracking (input/output)
• Model and provider information
• Error tracking with finish reasons
• Conforms to OpenTelemetry GenAI semantic conventions v1.37+

OTLP-Compatible Backends:

The tracing implementation uses standard OTLP HTTP protocol, so it works with any OTLP-compatible backend:

• Grafana LGTM (Recommended for local dev): All-in-one stack with Tempo traces, Loki logs, Mimir metrics, and Grafana UI
• Langfuse: LLM-focused observability and analytics
• OpenLIT: Built-in LLM dashboards, cost tracking
• DataDog, New Relic, Honeycomb: Commercial platforms

Example Configuration:

# Enable tracing
export HINDSIGHT_API_OTEL_TRACES_ENABLED=true

# Configure endpoint (example: OpenLIT Cloud)
export HINDSIGHT_API_OTEL_EXPORTER_OTLP_ENDPOINT=https://otlp.openlit.io
export HINDSIGHT_API_OTEL_EXPORTER_OTLP_HEADERS="Authorization=Bearer olit-xxx"

# Optional: Custom service name and environment
export HINDSIGHT_API_OTEL_SERVICE_NAME=hindsight-production
export HINDSIGHT_API_OTEL_DEPLOYMENT_ENVIRONMENT=production

Local Development:

For local development, we recommend the Grafana LGTM stack which provides traces, metrics, and logs in a single container:

./scripts/dev/start-grafana.sh

See scripts/dev/grafana/README.md for detailed setup instructions.

Other options: See scripts/dev/openlit/README.md for OpenLIT or scripts/dev/jaeger/README.md for standalone Jaeger.

Metrics

Hindsight exposes Prometheus metrics at the /metrics endpoint, including:

• LLM call duration and token usage
• Operation duration (retain/recall/reflect)
• HTTP request metrics
• Database connection pool metrics

Metrics are always enabled and available at http://localhost:8888/metrics.

---

Control Plane

The Control Plane is the web UI for managing memory banks.

Variable	Description	Default
HINDSIGHT_CP_DATAPLANE_API_URL	URL of the API service	http://localhost:8888
NEXT_PUBLIC_BASE_PATH	Base path for Control Plane UI when behind reverse proxy (e.g., /hindsight)	"" (root)

# Point Control Plane to a remote API service
export HINDSIGHT_CP_DATAPLANE_API_URL=http://api.example.com:8888

Hierarchical Configuration

Hindsight supports per-bank configuration overrides through a hierarchical system: Global (env vars) → Tenant → Bank.

Type-Safe Config Access

To prevent accidentally using global defaults when bank-specific overrides exist, Hindsight enforces type-safe config access:

In Application Code:

from hindsight_api.config import get_config

# ✅ Access static (infrastructure) fields
config = get_config()
host = config.host  # OK - static field
port = config.port  # OK - static field

# ❌ Attempting to access bank-configurable fields raises an error
chunk_size = config.retain_chunk_size  # ConfigFieldAccessError!

Error Message:

ConfigFieldAccessError: Field 'retain_chunk_size' is bank-configurable and cannot
be accessed from global config. Use ConfigResolver.resolve_full_config(bank_id, context)
to get bank-specific config.

For Bank-Specific Config:

# Internal code that needs bank-specific settings
from hindsight_api.config_resolver import ConfigResolver

# Resolve full config for a specific bank
config = await config_resolver.resolve_full_config(bank_id, request_context)
chunk_size = config.retain_chunk_size  # ✅ Uses bank-specific value

This design prevents bugs where global defaults are used instead of bank overrides, making it impossible to make this mistake at compile/development time.

Security Model

Configuration fields are categorized for security:

1. Configurable Fields - Safe behavioral settings that can be customized per-bank:

   • Retention: retain_chunk_size, retain_extraction_mode, retain_custom_instructions
   • Consolidation: enable_observations

2. Credential Fields - NEVER exposed or configurable via API:

   • API keys: *_api_key (all LLM API keys)
   • Infrastructure: *_base_url (all base URLs)

3. Static Fields - Server-level only, cannot be overridden:

   • Infrastructure: database_url, port, host, worker_count
   • Provider/Model selection: llm_provider, llm_model (requires presets - not yet implemented)
   • Performance tuning: llm_max_concurrent, llm_timeout, retrieval settings, optimization flags

Enabling the API

Variable	Description	Default
HINDSIGHT_API_ENABLE_BANK_CONFIG_API	Enable per-bank config API	false

Important: The bank config API is disabled by default for security. Enable it explicitly:

export HINDSIGHT_API_ENABLE_BANK_CONFIG_API=true

API Endpoints

• GET /v1/default/banks/{bank_id}/config - View resolved config (filtered by permissions)
• PATCH /v1/default/banks/{bank_id}/config - Update bank overrides (only allowed fields)
• DELETE /v1/default/banks/{bank_id}/config - Reset to defaults

Permission System

Tenant extensions can control which fields banks are allowed to modify via get_allowed_config_fields():

class CustomTenantExtension(TenantExtension):
    async def get_allowed_config_fields(self, context, bank_id):
        # Option 1: Allow all configurable fields
        return None

        # Option 2: Allow specific fields only
        return {"retain_chunk_size", "retain_custom_instructions"}

        # Option 3: Read-only (no modifications)
        return set()

Examples

# Update retention settings for a bank
curl -X PATCH http://localhost:8888/v1/default/banks/my-bank/config \
  -H "Content-Type: application/json" \
  -d '{
    "updates": {
      "retain_chunk_size": 4000,
      "retain_extraction_mode": "custom",
      "retain_custom_instructions": "Focus on technical details and implementation specifics"
    }
  }'

# Note: retain_extraction_mode must be "custom" to use retain_custom_instructions

# View resolved config (respects permissions)
curl http://localhost:8888/v1/default/banks/my-bank/config

# Reset to defaults
curl -X DELETE http://localhost:8888/v1/default/banks/my-bank/config

Security Notes:

• Credentials (API keys, base URLs) are never returned in responses
• Only configurable fields can be modified
• Responses are filtered by tenant permissions
• Attempting to set credentials returns 400 error

Reverse Proxy / Subpath Deployment

To deploy Hindsight under a subpath (e.g., example.com/hindsight/):

1. Set both environment variables to the same path:

   HINDSIGHT_API_BASE_PATH=/hindsight
   NEXT_PUBLIC_BASE_PATH=/hindsight
   

2. Configure your reverse proxy to:

   • Forward /hindsight/* requests to Hindsight
   • Preserve the full path in forwarded requests
   • Set appropriate proxy headers (X-Forwarded-Proto, X-Forwarded-For)

Example: Nginx Configuration

location /hindsight/ {
    proxy_pass http://localhost:8888/;
    proxy_set_header Host $host;
    proxy_set_header X-Real-IP $remote_addr;
    proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
    proxy_set_header X-Forwarded-Proto $scheme;
}

Example: Traefik Configuration

http:
  routers:
    hindsight:
      rule: "PathPrefix(`/hindsight`)"
      service: hindsight
      middlewares:
        - hindsight-stripprefix

  middlewares:
    hindsight-stripprefix:
      stripPrefix:
        prefixes:
          - "/hindsight"

  services:
    hindsight:
      loadBalancer:
        servers:
          - url: "http://localhost:8888"

Important Notes:

• The base path must start with / and should NOT end with /
• Both API and Control Plane should use the same base path
• After setting environment variables, restart both services
• OpenAPI docs will be available at <base-path>/docs (e.g., /hindsight/docs)

Complete Examples:

See docker/compose-examples/ directory for:

• Nginx configuration files (simple.conf, api-and-control-plane.conf)
• Docker Compose setups (docker-compose.yml, reverse-proxy-only.yml)
• Traefik and other reverse proxy examples
• Full deployment documentation

---

Example .env File

# API Service
HINDSIGHT_API_DATABASE_URL=postgresql://hindsight:hindsight_dev@localhost:5432/hindsight
# HINDSIGHT_API_DATABASE_SCHEMA=public  # optional, defaults to 'public'
HINDSIGHT_API_LLM_PROVIDER=groq
HINDSIGHT_API_LLM_API_KEY=gsk_xxxxxxxxxxxx

# Authentication (optional, recommended for production)
# HINDSIGHT_API_TENANT_EXTENSION=hindsight_api.extensions.builtin.tenant:ApiKeyTenantExtension
# HINDSIGHT_API_TENANT_API_KEY=your-secret-api-key

# Control Plane
HINDSIGHT_CP_DATAPLANE_API_URL=http://localhost:8888

---

For configuration issues not covered here, please open an issue on GitHub.