lethe-core-rust 0.1.1

High-performance hybrid retrieval engine combining BM25 lexical search with vector similarity using z-score fusion. Features hero configuration for optimal parity with splade baseline, gamma boosting for code/error contexts, and comprehensive chunking pipeline.
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48

# Example configuration for Lethe with proxy enabled
# Copy this to config.toml and adjust settings as needed

[app]
host = "0.0.0.0"
port = 3000
cors_origin = "*"

[database]
# Set DATABASE_URL environment variable instead
# url = "postgresql://user:password@localhost/lethe"

# Proxy configuration - enable transparent reverse proxy
[proxy]
enabled = true

# OpenAI provider configuration
[proxy.openai]
base_url = "https://api.openai.com"

# Anthropic provider configuration  
[proxy.anthropic]
base_url = "https://api.anthropic.com"

# Authentication mode: "passthrough" or "inject"
[proxy.auth]
mode = "passthrough"  # Use "inject" to inject API keys automatically

# API keys for inject mode (set via environment variables)
[proxy.auth.inject]
# openai_api_key = "${OPENAI_API_KEY}"      # Set OPENAI_API_KEY env var
# anthropic_api_key = "${ANTHROPIC_API_KEY}" # Set ANTHROPIC_API_KEY env var

# Request rewriting configuration
[proxy.rewrite]
enabled = true
max_request_bytes = 2_000_000
prelude_system = "You are a helpful AI assistant designed to provide accurate and helpful information."

# Security configuration
[proxy.security]
allowed_providers = ["openai", "anthropic"]
max_request_size_bytes = 10_000_000

# Proxy timeout configuration
[proxy.timeouts]
connect_ms = 5000
read_ms = 30000