Configuration Reference
All environment variables for Remembra.
Required
| Variable |
Description |
Example |
OPENAI_API_KEY |
OpenAI API key |
sk-... |
Server
| Variable |
Default |
Description |
REMEMBRA_HOST |
0.0.0.0 |
Server bind address |
REMEMBRA_PORT |
8787 |
Server port |
REMEMBRA_WORKERS |
1 |
Number of worker processes |
REMEMBRA_LOG_LEVEL |
INFO |
Logging level |
Database
| Variable |
Default |
Description |
REMEMBRA_DATABASE_PATH |
./remembra.db |
SQLite database path |
QDRANT_HOST |
localhost |
Qdrant server host |
QDRANT_PORT |
6333 |
Qdrant server port |
QDRANT_API_KEY |
- |
Qdrant API key (if secured) |
QDRANT_COLLECTION |
remembra |
Qdrant collection name |
Embeddings
| Variable |
Default |
Description |
REMEMBRA_EMBEDDING_PROVIDER |
openai |
Provider: openai, ollama, cohere |
REMEMBRA_EMBEDDING_MODEL |
text-embedding-3-small |
Model name |
REMEMBRA_EMBEDDING_DIMENSIONS |
1536 |
Vector dimensions |
OLLAMA_BASE_URL |
http://localhost:11434 |
Ollama server URL |
COHERE_API_KEY |
- |
Cohere API key |
| Variable |
Default |
Description |
REMEMBRA_SMART_EXTRACTION_ENABLED |
true |
Enable LLM extraction |
REMEMBRA_EXTRACTION_MODEL |
gpt-4o-mini |
Model for extraction |
REMEMBRA_EXTRACTION_TEMPERATURE |
0.0 |
Extraction temperature |
Entity Resolution
| Variable |
Default |
Description |
REMEMBRA_ENTITY_EXTRACTION_ENABLED |
true |
Extract entities |
REMEMBRA_ENTITY_MATCHING_THRESHOLD |
0.85 |
Alias matching threshold |
Retrieval
| Variable |
Default |
Description |
REMEMBRA_DEFAULT_THRESHOLD |
0.40 |
Default similarity threshold |
REMEMBRA_DEFAULT_LIMIT |
10 |
Default recall limit |
REMEMBRA_DEFAULT_MAX_TOKENS |
4000 |
Max context tokens |
Hybrid Search
| Variable |
Default |
Description |
REMEMBRA_HYBRID_SEARCH_ENABLED |
true |
Enable hybrid search |
REMEMBRA_HYBRID_ALPHA |
0.4 |
Keyword weight (0-1) |
REMEMBRA_HYBRID_FUSION |
weighted |
Fusion: weighted or rrf |
Reranking
| Variable |
Default |
Description |
REMEMBRA_RERANK_ENABLED |
false |
Enable CrossEncoder reranking |
REMEMBRA_RERANK_MODEL |
cross-encoder/ms-marco-MiniLM-L-6-v2 |
Reranker model |
REMEMBRA_RERANK_TOP_K |
20 |
Candidates to rerank |
Ranking Weights
| Variable |
Default |
Description |
REMEMBRA_RANKING_SEMANTIC_WEIGHT |
0.6 |
Semantic score weight |
REMEMBRA_RANKING_RECENCY_WEIGHT |
0.15 |
Recency boost weight |
REMEMBRA_RANKING_ENTITY_WEIGHT |
0.15 |
Entity match weight |
REMEMBRA_RANKING_KEYWORD_WEIGHT |
0.1 |
Keyword match weight |
REMEMBRA_RANKING_RECENCY_DECAY_DAYS |
30 |
Recency half-life (days) |
Graph Retrieval
| Variable |
Default |
Description |
REMEMBRA_GRAPH_RETRIEVAL_ENABLED |
true |
Enable graph traversal |
REMEMBRA_GRAPH_TRAVERSAL_DEPTH |
2 |
Max hop depth |
Temporal
| Variable |
Default |
Description |
REMEMBRA_DEFAULT_TTL_DAYS |
- |
Default TTL (optional) |
REMEMBRA_DECAY_ENABLED |
true |
Enable memory decay |
REMEMBRA_DECAY_HALF_LIFE_DAYS |
30 |
Decay half-life |
REMEMBRA_ACCESS_BOOST_WEIGHT |
0.2 |
Access count boost |
Security
Authentication
| Variable |
Default |
Description |
REMEMBRA_AUTH_ENABLED |
true |
Enable API key auth |
REMEMBRA_AUTH_MASTER_KEY |
- |
Master admin key |
Rate Limiting
| Variable |
Default |
Description |
REMEMBRA_RATE_LIMIT_ENABLED |
true |
Enable rate limiting |
REMEMBRA_RATE_LIMIT_STORAGE |
memory |
Backend: memory or redis://... |
REMEMBRA_RATE_LIMIT_STORE |
30/minute |
Store endpoint limit |
REMEMBRA_RATE_LIMIT_RECALL |
60/minute |
Recall endpoint limit |
REMEMBRA_RATE_LIMIT_FORGET |
10/minute |
Forget endpoint limit |
Sanitization
| Variable |
Default |
Description |
REMEMBRA_SANITIZATION_ENABLED |
true |
Enable input sanitization |
REMEMBRA_TRUST_SCORE_THRESHOLD |
0.5 |
Suspicious content threshold |
Dashboard
| Variable |
Default |
Description |
REMEMBRA_STATIC_DIR |
- |
Path to dashboard build |
Example .env File
# Required
OPENAI_API_KEY=sk-your-key-here
# Server
REMEMBRA_HOST=0.0.0.0
REMEMBRA_PORT=8787
# Database
REMEMBRA_DATABASE_PATH=/app/data/remembra.db
QDRANT_HOST=qdrant
QDRANT_PORT=6333
# Security (enable in production!)
REMEMBRA_AUTH_ENABLED=true
REMEMBRA_AUTH_MASTER_KEY=your-secure-master-key
REMEMBRA_RATE_LIMIT_ENABLED=true
# Extraction
REMEMBRA_SMART_EXTRACTION_ENABLED=true
REMEMBRA_EXTRACTION_MODEL=gpt-4o-mini
# Retrieval
REMEMBRA_HYBRID_SEARCH_ENABLED=true
REMEMBRA_RERANK_ENABLED=false
REMEMBRA_DEFAULT_MAX_TOKENS=4000
# Temporal
REMEMBRA_DEFAULT_TTL_DAYS=365
REMEMBRA_DECAY_ENABLED=true