Skip to main content

​
Overview

The CLI interface provides direct command-line access to RepoMaster’s multi-agent system. It’s ideal for automation, scripting, and users who prefer terminal-based workflows.

​
Available Commands

​
Basic Help

python launcher.py --help

​
Command Structure

All RepoMaster commands follow this pattern: 
python launcher.py --mode <MODE> [OPTIONS]

​
Primary Modes

From launcher.py:6-12, RepoMaster supports two primary modes:
ModeDescriptionUse Case
frontendInteractive Multi-Agent DashboardWeb-based visual interface
backendMulti-Agent Service InterfaceCLI-based agent interaction

​
Global Options

​
Configuration Flags

From launcher.py:478-483 and the argument parser:
FlagTypeDefaultDescription
--modestringrequiredExecution mode (frontend/backend)
--api-typestringbasicAPI provider (openai, claude, deepseek, etc.)
--temperaturefloat0.1Model temperature (0.0-2.0)
--work-dirpathautoWorking directory for code execution
--log-levelstringINFOLogging level (DEBUG, INFO, WARNING, ERROR)
--skip-config-checkflagfalseSkip API configuration validation

​
Frontend-Specific Options

FlagTypeDefaultDescription
--streamlit-portint8501Web server port
--streamlit-hoststringlocalhostServer host address
--max-upload-sizeint200Upload limit in MB (1-2000)
--file-watcher-typestringautoFile watching mechanism

​
Backend-Specific Options

FlagTypeDefaultDescription
--backend-modestringrequiredAgent mode (unified, deepsearch, etc.)
--timeoutint120Request timeout in seconds
--max-tokensint4000Maximum token count

​
Usage Examples

​
Frontend Mode

python launcher.py --mode frontend

​
Backend Mode

python launcher.py --mode backend --backend-mode unified

​
Advanced Configuration

python launcher.py --mode backend \
  --backend-mode unified \
  --api-type claude \
  --temperature 0.7

​
Interactive Commands

When running in backend mode, the CLI provides interactive commands:

​
Conversation Control

From launcher.py:189-215 and similar sections:
CommandAliasesDescription
quitexit, qExit the program
historyhView conversation history
clearcClear conversation context

​
Example Session

$ python launcher.py --mode backend --backend-mode unified

🤖 RepoMaster Multi-Agent System Ready!

🤖 Please describe your task: Analyze the FastAPI repository structure
🔧 Intelligent task analysis...
   📊 Selecting optimal processing method...

📋 Task execution result:
[Agent analysis output...]

🤖 Please describe your task: history
[Conversation history displayed...]

🤖 Please describe your task: clear
Conversation history cleared.

🤖 Please describe your task: quit
👋 Multi-Agent system service stopped

​
Environment Setup

​
Prerequisites Check

From launcher.py:50-113, the launcher automatically checks:
  1. Environment file (configs/.env)
  2. Required API keys (SERPER_API_KEY, JINA_API_KEY)
  3. PYTHONPATH configuration
The launcher will exit with helpful error messages if:
  • Configuration file is missing (suggests copying from env.example)
  • Required API keys are not set
  • Configuration validation fails (unless --skip-config-check is used)

​
Configuration File Setup

If the .env file is missing, you’ll see: 
⚠️  Configuration file not found!
📝 Please copy the example configuration file:
   cp configs/env.example configs/.env
   Then edit configs/.env with your API keys
💡 See README.md or USAGE.md for detailed setup instructions

​
Missing API Keys

If required keys are missing: 
⚠️  Missing required API keys in .env file: SERPER_API_KEY, JINA_API_KEY
📝 Please edit configs/.env and add the missing keys
💡 See README.md or USAGE.md for API key setup instructions

​
Logging and Debugging

​
Log Levels

From launcher.py:38-48, available log levels:
LevelUsageOutput
DEBUGDevelopment, troubleshootingAll messages including library internals
INFONormal operationStatus updates and important events
WARNINGProduction (default)Warnings and errors only
ERRORMinimal outputErrors only

​
Enable Debug Logging

python launcher.py --mode backend \
  --backend-mode unified \
  --log-level DEBUG
Debug logging is automatically filtered for noisy third-party libraries:
  • autogen.oai.client - Set to ERROR level
  • langchain_community.utils.user_agent - Set to ERROR level

​
Startup Sequence

From launcher.py:490-580, the launcher follows this sequence:
  1. Display CLI logo (optional)
  2. Setup environment
    • Configure PYTHONPATH
    • Load .env file
    • Validate API keys
  3. Parse command-line arguments
  4. Configure logging
  5. Validate API configuration (unless skipped)
  6. Create configuration manager
  7. Print startup panel with status
  8. Launch selected mode

​
Error Handling

​
Common Errors

​
Configuration Errors

❌ Startup failed: Unsupported running mode: invalid_mode
Solution: Use --mode frontend or --mode backend

​
Backend Mode Missing

❌ Startup failed: Unsupported backend mode: invalid_backend
Solution: Specify valid --backend-mode (unified, deepsearch, general_assistant, repository_agent)

​
Environment Issues

❌ Missing required environment variables: SERPER_API_KEY, JINA_API_KEY
Solution: Configure configs/.env with required API keys

​
Graceful Shutdown

From launcher.py:573-574, the launcher handles keyboard interrupts: 
^C
👋 Program interrupted by user
Agent-specific shutdowns:
  • Deep Search: 👋 Deep Search Agent service stopped (launcher.py:215)
  • Programming Assistant: 👋 Programming Assistant Agent service stopped (launcher.py:284)
  • Repository Agent: 👋 Repository Exploration Agent service stopped (launcher.py:370)
  • Unified Mode: 👋 Multi-Agent system service stopped (launcher.py:435)

​
Shell Script Shortcuts

RepoMaster includes a run.sh script for convenient access: 
bash run.sh frontend

​
Best Practices

​
Development Workflow

For development:
python launcher.py --mode backend \
  --backend-mode unified \
  --log-level DEBUG \
  --work-dir ./dev-workspace

​
Production Deployment

For production:
python launcher.py --mode frontend \
  --streamlit-host 0.0.0.0 \
  --streamlit-port 8501 \
  --api-type openai \
  --log-level WARNING

​
Automated Scripts

For automation, use --skip-config-check to avoid interactive prompts:
python launcher.py --mode backend \
  --backend-mode unified \
  --skip-config-check

​
Performance Tuning

​
Model Temperature

Adjust creativity vs consistency: 
# More deterministic (recommended for code)
python launcher.py --mode backend \
  --backend-mode unified \
  --temperature 0.1

# More creative (better for brainstorming)
python launcher.py --mode backend \
  --backend-mode unified \
  --temperature 0.7

​
Request Timeouts

# Longer timeout for complex tasks
python launcher.py --mode backend \
  --backend-mode unified \
  --timeout 300

​
Token Limits

# Increase for longer conversations
python launcher.py --mode backend \
  --backend-mode unified \
  --max-tokens 8000

​
See Also

Build docs developers (and LLMs) love

Get started for freeTalk to us