-
Notifications
You must be signed in to change notification settings - Fork 3
Expand file tree
/
Copy path.env.example
More file actions
54 lines (46 loc) · 2.31 KB
/
Copy path.env.example
File metadata and controls
54 lines (46 loc) · 2.31 KB
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
52
53
54
## Common
AUTH_OIDC_ISSUER="http://localhost:8080"
## Backend
OPENAI_API_KEY=
# Under `docker compose`, a host-local Ollama (or other provider bound to the
# Docker host) must be addressed as http://host.docker.internal:11434/v1 —
# `localhost` from inside the container means the container itself.
OPENAI_BASE_URL=
# Use a "{provider}:{model}" prefix (e.g. "openrouter:deepseek/deepseek-r1") to route
# through a specific langchain-supported provider integration instead of the generic
# OpenAI-compatible path above.
CHAT_MODEL_NAME=gpt-4o-mini
# Model used to generate thread titles. Optional -- falls back to CHAT_MODEL_NAME.
# When set, CHAT_MODEL_KWARGS (provider-specific to the chat model) is not applied to it.
# TITLE_MODEL_NAME=gpt-4o-mini
# PostgreSQL connection for Aegra. Leave unset for the common setups:
# - `moon backend:dev` on the host -> Aegra defaults to localhost:5432/aegra,
# which is what `moon backend:docker-postgres` serves
# - `docker compose up` -> compose points the backend at the `postgres` service
# Set it only for your own/external server, and make sure the host is reachable
# from wherever the backend runs (inside the container, `localhost` is the
# container itself — use `host.docker.internal` or a real hostname).
# DATABASE_URL=postgresql://user:pass@db.example.com:5432/aegra
# E2E uses a separate TEST_DATABASE_URL, defined in .env.e2e, which must name
# a different database — the reset script drops and recreates it on every run.
AUTH_TYPE=custom
### Optional observability (OpenTelemetry fan-out, e.g. Langfuse or Phoenix)
# OTEL_TARGETS=
# LANGFUSE_BASE_URL=
# LANGFUSE_PUBLIC_KEY=
# LANGFUSE_SECRET_KEY=
# Extra kwargs passed to init_chat_model as JSON, e.g. to request reasoning tokens (opt-in, required for OpenRouter); quote and escape inner double quotes so .env parsing doesn't mangle them:
# CHAT_MODEL_KWARGS="{\"reasoning\":{\"effort\":\"low\"}}"
# Required when using the "openrouter:" model provider prefix:
# OPENROUTER_API_KEY=
# OPENROUTER_API_BASE=https://openrouter.ai/api/v1
## Frontend
AUTH_TRUST_HOST="true"
### Mock OIDC server
AUTH_OIDC_CLIENT_ID="svelte-langgraph"
AUTH_OIDC_CLIENT_SECRET="secret"
### Generate with `npx auth`. Read more: https://cli.authjs.dev
AUTH_SECRET=
#### Default host for backend
PUBLIC_LANGGRAPH_API_URL="http://localhost:2026"
PUBLIC_SENTRY_DSN=""