varynth 0.1.0

Varynth CLI — OpenClaw-style coding agent with a local dashboard
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

model = "grok-4.7-build"

provider = "proxy"

permission_mode = "acceptEdits"

sandbox = "workspace-write"

theme = "purple"

# use_keyring = true

# docker_image = "varynth-sandbox:latest"

# docker_network = "none"

proxy_url = "http://127.0.0.1:8787"

# proxy_token = "set via VARYNTH_PROXY_TOKEN"

# OpenAI-compatible endpoint (models: grok-4.5, grok-4.7-build, grok-4.7, grok-4.7-build):

# provider = "openai"

# model = "grok-4.7-build"

# openai_base_url = "https://codex-everywhere.com"

# openai_api_key = "set via OPENAI_API_KEY"

# Direct Anthropic endpoint:

# provider = "anthropic"

# anthropic_base_url = "https://api.anthropic.com"

# anthropic_api_key = "set via ANTHROPIC_API_KEY"

# effort = "xhigh"  # one of: low, medium, high, xhigh, max, ultra

# max_tokens = 8192  # optional cap on model output tokens

# temperature = 0.7  # optional sampling temperature

# goal_max_rounds = 25  # optional cap for /goal loops

# goal_max_minutes = 30  # optional wall-time cap for /goal

# stream = true  # set false to disable SSE token streaming

# add_dirs = ["C:/path/to/extra/workspace"]

dashboard_host = "127.0.0.1"

dashboard_port = 7420

# dashboard_token = "set via VARYNTH_DASHBOARD_TOKEN for remote API access"

# mcp_config = "C:/path/to/mcp.json"

# Telegram bot (or set VARYNTH_TELEGRAM_BOT_TOKEN / telegram.env):

# telegram_bot_token = ""

telegram_allow_from = ["123456789"]

# GitHub + Google connectors read env vars only; no config keys for them:

#   VARYNTH_GITHUB_TOKEN, VARYNTH_GITHUB_ACCOUNT

#   VARYNTH_GOOGLE_TOKEN, VARYNTH_GOOGLE_ACCOUNT

max_tool_rounds = 24

shell_allowlist = [

  "git", "cargo", "rustc", "npm", "npx", "pnpm", "node",

  "python", "rg", "powershell", "pwsh", "cmd"

]