Repository navigation
Expand file tree
/
Copy path.env.example
More file actions
47 lines (41 loc) · 2.05 KB
/
Copy path.env.example
File metadata and controls
47 lines (41 loc) · 2.05 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
# SPDX-FileCopyrightText: Sudo Apt Holdings LLC
# SPDX-License-Identifier: Apache-2.0
#
# Copy to .env and fill in the values. .env is ignored by git (.gitignore, checked by
# scripts/plan_check.sh rule 9); this example file is the only one that is tracked.
#
# Load into a shell before running live tests or the app:
#
# set -a; . ./.env; set +a
#
# The default test run blocks the network (test/support/network_guard.ex). Anything that
# talks to a real provider is tagged :live and runs only with `mix test --only live`.
# Keys never go in config files, commit messages, PROOF.md or NOTES.md.
# ---- LLM providers for live tests (slice 011 onward) ------------------------------------
# One hosted provider with a free tier is enough for the live suite. Both below speak the
# OpenAI-compatible API, which req_llm drives through its openai_compatible provider.
# OpenRouter: https://openrouter.ai/keys
OPENROUTER_API_KEY=
OPENROUTER_BASE_URL=https://openrouter.ai/api/v1
# NVIDIA NIM: https://build.nvidia.com/ (API key from the account page). Named after the
# model family it is used for. Model ids there reach end of life: the id you set is checked
# against GET /models before the live suite runs, and a 410 names the replacement to find.
NEMOTRON_API_KEY=
NEMOTRON_BASE_URL=https://integrate.api.nvidia.com/v1
NEMOTRON_MODEL=
# Which provider and model the live suite uses. The model id must exist on the provider's
# current list and answer a tool call and a streamed completion; check both there rather
# than here, since free-tier ids change. The pick is recorded in the slice's NOTES.md with
# the probe output.
TRINITY_LIVE_PROVIDER=openrouter
TRINITY_LIVE_MODEL=
# ---- Local inference, optional --------------------------------------------------------
# Ollama or LM Studio through the same openai_compatible provider. Leave empty to skip.
OLLAMA_BASE_URL=http://localhost:11434/v1
OLLAMA_MODEL=
# ---- App runtime (config/runtime.exs reads these; all have defaults) ------------------
# PORT=4000
# DATABASE_PATH=
# SECRET_KEY_BASE=
# PHX_HOST=
# TRINITY_AUTHORITY=local