forked from Saik0s/mcp-browser-use
-
Notifications
You must be signed in to change notification settings - Fork 0
Expand file tree
/
Copy path.env.example
More file actions
145 lines (125 loc) · 6.4 KB
/
Copy path.env.example
File metadata and controls
145 lines (125 loc) · 6.4 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
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
# === LLM Provider Configuration ===
# Select the primary LLM provider for agent tasks
# Options: openai, azure_openai, anthropic, google, mistral, ollama, deepseek, openrouter, alibaba, moonshot, unbound
MCP_MODEL_PROVIDER=openrouter
# Specify the model name for the selected provider (e.g., gpt-4o, claude-3-7-sonnet-20250219, gemini-1.5-flash-latest)
MCP_MODEL_NAME=anthropic/claude-3.7-sonnet
# Optional: Generic override for the LLM provider's base URL
# MCP_BASE_URL=
# Optional: Generic override for the LLM provider's API key. Takes precedence over provider-specific keys.
# MCP_API_KEY=
# --- Provider Specific API Keys & Endpoints ---
# Required unless using Ollama locally without auth or generic MCP_API_KEY is set
# OpenAI
OPENAI_API_KEY=YOUR_OPENAI_API_KEY
# OPENAI_ENDPOINT=https://api.openai.com/v1 # Optional: Override default endpoint
# Anthropic
ANTHROPIC_API_KEY=YOUR_ANTHROPIC_API_KEY
# ANTHROPIC_ENDPOINT=https://api.anthropic.com # Optional: Override default endpoint
# Google (Gemini)
GOOGLE_API_KEY=YOUR_GOOGLE_API_KEY
# Azure OpenAI
AZURE_OPENAI_API_KEY=YOUR_AZURE_OPENAI_API_KEY
AZURE_OPENAI_ENDPOINT=YOUR_AZURE_ENDPOINT # Required: e.g., https://your-resource.openai.azure.com/
AZURE_OPENAI_API_VERSION=2025-01-01-preview # Optional: Specify API version
# DeepSeek
DEEPSEEK_API_KEY=YOUR_DEEPSEEK_API_KEY
# DEEPSEEK_ENDPOINT=https://api.deepseek.com # Optional: Override default endpoint
# Mistral
MISTRAL_API_KEY=YOUR_MISTRAL_API_KEY
# MISTRAL_ENDPOINT=https://api.mistral.ai/v1 # Optional: Override default endpoint
# Ollama (Usually local, no key needed by default)
OLLAMA_ENDPOINT=http://localhost:11434 # Optional: Override default endpoint if Ollama runs elsewhere
# OLLAMA_NUM_CTX=32000 # Optional: Context window size for Ollama models
# OLLAMA_NUM_PREDICT=1024 # Optional: Max tokens to predict for Ollama models
# OpenRouter
OPENROUTER_API_KEY=YOUR_OPENROUTER_API_KEY
# OPENROUTER_ENDPOINT=https://openrouter.ai/api/v1 # Optional: Override default endpoint
# Alibaba (DashScope)
ALIBABA_API_KEY=YOUR_ALIBABA_API_KEY
# ALIBABA_ENDPOINT=https://dashscope.aliyuncs.com/compatible-mode/v1 # Optional: Override default endpoint
# Moonshot
MOONSHOT_API_KEY=YOUR_MOONSHOT_API_KEY
# MOONSHOT_ENDPOINT=https://api.moonshot.cn/v1 # Optional: Override default endpoint
# Unbound AI
UNBOUND_API_KEY=YOUR_UNBOUND_API_KEY
# UNBOUND_ENDPOINT=https://api.getunbound.ai # Optional: Override default endpoint
# --- LLM Behavior ---
# Controls randomness (0.0 = deterministic, >1.0 = more random)
MCP_TEMPERATURE=0.0
# Tool calling method ('auto', 'json_schema', 'function_calling') - affects how LLM interacts with tools
MCP_TOOL_CALLING_METHOD=auto
# Maximum input tokens allowed for the LLM context
MCP_MAX_INPUT_TOKENS=128000
# === Agent Configuration (`run_browser_agent` tool) ===
# Agent implementation to use ('org' or 'custom')
MCP_AGENT_TYPE=org
# Maximum number of steps an agent run can take
MCP_MAX_STEPS=100
# Enable vision capabilities for the agent (screenshot analysis)
MCP_USE_VISION=true
# Maximum number of actions the agent can perform in a single step
MCP_MAX_ACTIONS_PER_STEP=5
# Keep browser managed by server open between agent runs (only applies if MCP_USE_OWN_BROWSER=false)
MCP_KEEP_BROWSER_OPEN=false
# Enable Playwright video recording for agent runs (true/false)
MCP_ENABLE_RECORDING=false
# Optional: Path to save agent run video recordings (e.g., ./tmp/recordings). Required if MCP_ENABLE_RECORDING=true.
# MCP_SAVE_RECORDING_PATH=
# Path to save agent history JSON files
MCP_AGENT_HISTORY_PATH=./tmp/agent_history
# Run browser without UI specifically for `run_browser_agent` tool (true/false)
MCP_HEADLESS=true
# Disable browser security features specifically for `run_browser_agent` tool (use cautiously) (true/false)
MCP_DISABLE_SECURITY=true
# === Deep Research Configuration (`run_deep_search` tool) ===
# Maximum search iterations for deep research tasks
MCP_RESEARCH_MAX_ITERATIONS=10
# Maximum search queries per iteration in deep research
MCP_RESEARCH_MAX_QUERY=3
# Use a separate browser instance for deep research (true/false)
MCP_RESEARCH_USE_OWN_BROWSER=false
# Optional: Directory to save deep research artifacts (reports, intermediate results)
# MCP_RESEARCH_SAVE_DIR=./tmp/deep_research_results
# Maximum steps for sub-agents used within deep research
MCP_RESEARCH_AGENT_MAX_STEPS=10
# === Browser Configuration (General & Specific Tool Overrides) ===
# Set to true to connect to user's browser via CHROME_CDP instead of launching a new one
MCP_USE_OWN_BROWSER=false
# Optional: Connect to an existing Chrome instance via Chrome DevTools Protocol URL (e.g., http://localhost:9222). Required if MCP_USE_OWN_BROWSER=true.
# CHROME_CDP=http://localhost:9222
# Run the browser without a visible UI (true/false). Primarily affects `run_deep_search`. See also MCP_HEADLESS for `run_browser_agent`.
BROWSER_HEADLESS=true
# General browser security setting (true/false). See also MCP_DISABLE_SECURITY for `run_browser_agent`.
BROWSER_DISABLE_SECURITY=false
# Optional: Path to Chrome/Chromium executable if not in default location
# CHROME_PATH=
# Optional: Path to a Chrome user data directory for persistent sessions/profiles (useful with CHROME_CDP)
# CHROME_USER_DATA=
# Path to save Playwright trace files (useful for debugging)
BROWSER_TRACE_PATH=./tmp/trace
# Browser window width in pixels
BROWSER_WINDOW_WIDTH=1280
# Browser window height in pixels
BROWSER_WINDOW_HEIGHT=720
# === Server & Logging ===
# Path for the server log file
LOG_FILE=mcp_server_browser_use.log
# Logging level: DEBUG, INFO, WARNING, ERROR, CRITICAL
BROWSER_USE_LOGGING_LEVEL=INFO
# Set to false to disable anonymized telemetry
ANONYMIZED_TELEMETRY=true
# === Deprecated / Less Common ===
# These were present in previous versions or are less commonly used.
# CHROME_DEBUGGING_PORT=9222 # Use CHROME_CDP instead
# CHROME_DEBUGGING_HOST=localhost # Use CHROME_CDP instead
# CHROME_PERSISTENT_SESSION=false # Use CHROME_USER_DATA instead
# RESOLUTION=1920x1080x24 # Use BROWSER_WINDOW_WIDTH/HEIGHT
# RESOLUTION_WIDTH=1920 # Use BROWSER_WINDOW_WIDTH
# RESOLUTION_HEIGHT=1080 # Use BROWSER_WINDOW_HEIGHT
# VNC_PASSWORD=youvncpassword # VNC is not directly configured here
# PYTHONIOENCODING=utf-8 # Generally handled by Python 3
# PYTHONUNBUFFERED=1 # Useful for Docker logs, often set in Dockerfile/runtime
# PYTHONUTF8=1 # Generally handled by Python 3
# BROWSER_RECORDING_PATH= # Replaced by MCP_SAVE_RECORDING_PATH for agent runs
# MCP_GENERATE_GIF=false # Handled internally by agent now