serverless-rag-chatbot/env.example at main · certainly-param/serverless-rag-chatbot · GitHub

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
49
50
51
## Vercel AI SDK (Gemini) - Only used for CHAT, not embeddings
# https://ai-sdk.dev
GOOGLE_GENERATIVE_AI_API_KEY=

# Optional: Gemini model to use for chat (default: gemini-2.5-flash)
# Options: gemini-2.5-flash (default, free tier, 10 RPM/250 RPD), gemini-2.5-flash-lite (free tier, 15 RPM/1,000 RPD), gemini-2.0-flash-lite (free tier, 30 RPM), gemini-2.5-pro (best quality, higher cost)
# GEMINI_MODEL=gemini-2.5-flash

## Upstash Redis (REST) - For semantic caching
UPSTASH_REDIS_REST_URL=
UPSTASH_REDIS_REST_TOKEN=

## Upstash Vector (REST) - For document storage and retrieval
# IMPORTANT: Create an index with BUILT-IN EMBEDDING MODEL (e.g., BAAI/bge-small-en-v1.5)
# This avoids external embedding API costs - Upstash handles embeddings for free!
UPSTASH_VECTOR_REST_URL=
UPSTASH_VECTOR_REST_TOKEN=

## gRPC ingest gateway (recommended for production)
# Next.js `/api/ingest` calls this service when set. Format: host:port
# Local docker-compose: INGEST_GRPC_URL=ingest-grpc:50051 and INGEST_GRPC_TLS=0
# Cloud Run (TLS on 443): INGEST_GRPC_URL=your-service-xxx.run.app:443
# INGEST_GRPC_URL=
# INGEST_GRPC_TLS=1
# (optional) direct|none — if gRPC fails, fall back to direct Upstash upsert in Next.js
# INGEST_GRPC_FALLBACK=direct
# (optional) gRPC deadline in ms
# INGEST_GRPC_DEADLINE_MS=120000
# Service listens on PORT (Cloud Run) or INGEST_GRPC_PORT (local)
# INGEST_GRPC_PORT=50051

## Observability (optional)
# OpenTelemetry: export traces to any OTLP backend (Langfuse, Jaeger, Grafana, etc.)
# OTEL_SERVICE_NAME=serverless-rag
# OTEL_EXPORTER_OTLP_TRACES_ENDPOINT=https://your-otel-backend/v1
# Langfuse: LLM observability (traces, token usage, cost). Get keys at https://langfuse.com
# LANGFUSE_PUBLIC_KEY=
# LANGFUSE_SECRET_KEY=
# LANGFUSE_BASE_URL=https://cloud.langfuse.com

## Playwright E2E (optional)
# CI runs smoke tests against a local `next start` server (see playwright.config.ts).
# Full-stack tests: set E2E_FULL=1 and use real keys in the env for the Next server.
# E2E_FULL=1
# Optional path to a PDF for ingest E2E (default: e2e/fixtures/sample.pdf)
# E2E_FIXTURE_PDF=
# Point Playwright at an already-running app (skip auto webServer)
# PLAYWRIGHT_SKIP_WEB_SERVER=1
# Default E2E app URL is http://127.0.0.1:3100 (see playwright.config.ts)
# PLAYWRIGHT_E2E_PORT=3100
# PLAYWRIGHT_BASE_URL=http://127.0.0.1:3100