chatgpt-api/.env.example at main · suphotP/chatgpt-api · 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
52
53
54
55
56
57
# Default provider registered by the CLI.
CHAT_PROVIDER=chatgpt

# ChatGPT Web auth. Leave blank until you are ready to wire real tokens.
CHATGPT_ACCESS_TOKEN=
CHATGPT_PROOF_TOKEN=
CHATGPT_TURNSTILE_TOKEN=
CHATGPT_COOKIES_JSON={}
CHATGPT_HEADERS_JSON={}

# Optional HAR import source copied from a logged-in ChatGPT browser session.
CHATGPT_HAR_PATH=~/.chatgpt-api/captures/chatgpt.har

# Preferred for manual browser capture: paste the copied Network request summary
# for https://chatgpt.com/backend-api/f/conversation into this ignored local file.
CHATGPT_REQUEST_PATH=./secrets/chatgpt-request.txt

# Optional multi-account profile support. Captures live under:
# ./secrets/accounts/<account>/chatgpt-request.txt
CHATGPT_ACCOUNT=
CHATGPT_ACCOUNTS=
CHATGPT_ACCOUNT_STRATEGY=auto
CHATGPT_ACCOUNTS_DIR=./secrets/accounts

# Optional local OpenAI-shaped API server settings.
CHATGPT_API_HOST=127.0.0.1
CHATGPT_API_PORT=8000
CHATGPT_API_KEY=local-dev-key
BRIDGE_CONSOLE_PORT=8080
CHARACTER_GAME_PORT=3000
CHATGPT_CONSOLE_URL=http://127.0.0.1:8080
CHATGPT_CONSOLE_COMMAND=docker compose up -d bridge-console
CHATGPT_PUBLIC_BASE_URL=http://127.0.0.1:8000/v1
CHATGPT_WEB_TIMEOUT=5400
CHATGPT_IMAGE_OUTPUT_DIR=./outputs/chatgpt-images
CHATGPT_RESEARCH_OUTPUT_DIR=./outputs/chatgpt-research
CHATGPT_ADMIN_DB_PATH=./outputs/chatgpt-admin.sqlite

# Recommended local throttles. You can raise these, but ChatGPT Web may still
# apply hidden short rate limits if image/research jobs are fired too quickly.
CHATGPT_CHAT_CONCURRENCY=free=1,go=2,plus=3,pro=4
CHATGPT_UPLOAD_CONCURRENCY=free=1,go=1,plus=1,pro=1
CHATGPT_IMAGE_CONCURRENCY=free=1,go=1,plus=2,pro=3
CHATGPT_RESEARCH_CONCURRENCY=free=1,go=1,plus=2,pro=2

# Tool bridge prompt mode for agent clients:
# optimized = compact prompt for small/POC work
# opencode  = fuller prompt preserving opencode's long agent context and skills
CHATGPT_AGENT_MODE=optimized
CHATGPT_MODEL_FALLBACK=auto
CHATGPT_TEMPORARY_CHAT=true

# opencode launcher defaults. The launcher stores the same values in
# ~/.config/chatgpt-api/opencode-setup.json after setup.
CHATGPT_OPENCODE_MODEL=chatgpt-web/auto
CHATGPT_OPENCODE_AGENT_MODE=optimized
CHATGPT_OPENCODE_BASE_URL=http://127.0.0.1:8000/v1