Qwen CLI Agent - DEBUG tracing

Overview

QwenCodeCLIAgent supports detailed DEBUG logging for troubleshooting and monitoring qwen-code CLI execution. You get a full execution trace including:

• CLI commands and arguments
• Input (prompt)
• Output (result)
• Environment variables
• Timing
• Errors and warnings

Quick start

1. Configure logging

from pathlib import Path
from config.logging_config import setup_logging

# Enable DEBUG logging
setup_logging(
    log_level="DEBUG",
    log_file=Path("logs/qwen_debug.log"),
    enable_debug_trace=True
)

2. Use the agent

from src.agents.qwen_code_cli_agent import QwenCodeCLIAgent

# Create agent
agent = QwenCodeCLIAgent(
    enable_web_search=True,
    enable_git=True,
    timeout=300
)

# Process content
content = {
    "text": "Your text for processing",
    "urls": ["https://example.com"]
}

result = await agent.process(content)

3. Check logs

Logs will be written to:

• logs/qwen_debug.log — main log (all levels)
• logs/qwen_debug_debug.log — DEBUG-only messages

What gets logged

Agent initialization

[DEBUG] [QwenCodeCLIAgent._check_cli_available] Checking qwen CLI availability...
[DEBUG] [QwenCodeCLIAgent._check_cli_available] CLI path: qwen
[DEBUG] [QwenCodeCLIAgent._check_cli_available] Running command: qwen --version
[DEBUG] [QwenCodeCLIAgent._check_cli_available] Return code: 0
[DEBUG] [QwenCodeCLIAgent._check_cli_available] STDOUT: qwen version 1.0.0
[DEBUG] [QwenCodeCLIAgent._check_cli_available] STDERR:

CLI execution

[DEBUG] [QwenCodeCLIAgent._execute_qwen_cli] === CLI EXECUTION TRACE START ===
[DEBUG] [QwenCodeCLIAgent._execute_qwen_cli] Command: qwen
[DEBUG] [QwenCodeCLIAgent._execute_qwen_cli] Working dir: /workspace
[DEBUG] [QwenCodeCLIAgent._execute_qwen_cli] Environment variables:
[DEBUG] [QwenCodeCLIAgent._execute_qwen_cli]   OPENAI_API_KEY=sk-xxxxx...
[DEBUG] [QwenCodeCLIAgent._execute_qwen_cli]   OPENAI_BASE_URL=https://api.openai.com

[DEBUG] [QwenCodeCLIAgent._execute_qwen_cli] === STDIN (PROMPT) ===
[DEBUG] [QwenCodeCLIAgent._execute_qwen_cli] Prompt length: 1234 characters
[DEBUG] [QwenCodeCLIAgent._execute_qwen_cli] Prompt preview (first 500 chars):
You are an autonomous content processing agent...

[DEBUG] [QwenCodeCLIAgent._execute_qwen_cli] === FULL PROMPT ===
[DEBUG] [QwenCodeCLIAgent._execute_qwen_cli]
<full prompt text>
[DEBUG] [QwenCodeCLIAgent._execute_qwen_cli] === END FULL PROMPT ===

[DEBUG] [QwenCodeCLIAgent._execute_qwen_cli] Starting subprocess at 1234567890.123
[DEBUG] [QwenCodeCLIAgent._execute_qwen_cli] Process created with PID: 12345
[DEBUG] [QwenCodeCLIAgent._execute_qwen_cli] Sending prompt to stdin...

[DEBUG] [QwenCodeCLIAgent._execute_qwen_cli] Process completed in 15.42s
[DEBUG] [QwenCodeCLIAgent._execute_qwen_cli] Process return code: 0

[DEBUG] [QwenCodeCLIAgent._execute_qwen_cli] === STDERR OUTPUT ===
[DEBUG] [QwenCodeCLIAgent._execute_qwen_cli] STDERR length: 0 characters
[DEBUG] [QwenCodeCLIAgent._execute_qwen_cli] STDERR is empty
[DEBUG] [QwenCodeCLIAgent._execute_qwen_cli] === END STDERR OUTPUT ===

[DEBUG] [QwenCodeCLIAgent._execute_qwen_cli] === STDOUT OUTPUT ===
[DEBUG] [QwenCodeCLIAgent._execute_qwen_cli] STDOUT length: 5678 characters
[DEBUG] [QwenCodeCLIAgent._execute_qwen_cli] STDOUT preview (first 500 chars):
# Machine Learning Article
...

[DEBUG] [QwenCodeCLIAgent._execute_qwen_cli] === FULL STDOUT ===
[DEBUG] [QwenCodeCLIAgent._execute_qwen_cli]
<full result text>
[DEBUG] [QwenCodeCLIAgent._execute_qwen_cli] === END FULL STDOUT ===

[DEBUG] [QwenCodeCLIAgent._execute_qwen_cli] === CLI EXECUTION TRACE END ===

Result processing

[DEBUG] [QwenCodeCLIAgent] STEP 1: Preparing prompt for qwen-code
[DEBUG] [QwenCodeCLIAgent] Prepared prompt length: 1234 characters

[DEBUG] [QwenCodeCLIAgent] STEP 2: Executing qwen-code CLI
[DEBUG] [QwenCodeCLIAgent] Received result length: 5678 characters

[DEBUG] [QwenCodeCLIAgent] STEP 3: Parsing agent response with standard parser
[DEBUG] [QwenCodeCLIAgent] Result text preview (first 500 chars): ...
[DEBUG] [QwenCodeCLIAgent] Files created: ['path/to/file.md']
[DEBUG] [QwenCodeCLIAgent] Folders created: ['path/to/folder']

[INFO] [QwenCodeCLIAgent] Successfully processed content: title='Machine Learning'

Log levels

DEBUG

• Use in development
• Full CLI execution details

INFO

• Use in production
• Key events only

WARNING

• Minimal logging, warnings only

ERROR

• Critical errors only

Examples

Example 1: Debugging CLI issues

import asyncio
from pathlib import Path
from config.logging_config import setup_logging
from src.agents.qwen_code_cli_agent import QwenCodeCLIAgent

async def debug_cli_issue():
    # Enable detailed logging
    setup_logging(
        log_level="DEBUG",
        log_file=Path("logs/debug.log"),
        enable_debug_trace=True
    )

    agent = QwenCodeCLIAgent(timeout=60)

    try:
        result = await agent.process({"text": "Test content", "urls": []})
        print(f"Success: {result['title']}")
    except Exception as e:
        print(f"Error: {e}")
        print("Check logs/debug.log and logs/debug_debug.log for details")

asyncio.run(debug_cli_issue())

Example 2: Production monitoring

import asyncio
from pathlib import Path
from config.logging_config import setup_logging
from src.agents.qwen_code_cli_agent import QwenCodeCLIAgent

async def production_monitoring():
    # Moderate logging for production
    setup_logging(
        log_level="INFO",
        log_file=Path("logs/production.log"),
        enable_debug_trace=False
    )

    agent = QwenCodeCLIAgent()
    result = await agent.process(content)

    # Logs contain only important events (start/end, title/category, errors)

asyncio.run(production_monitoring())

Example 3: Performance analysis

import asyncio
import time
from pathlib import Path
from config.logging_config import setup_logging
from src.agents.qwen_code_cli_agent import QwenCodeCLIAgent

async def performance_analysis():
    # DEBUG for timing analysis
    setup_logging(
        log_level="DEBUG",
        log_file=Path("logs/performance.log"),
        enable_debug_trace=True
    )

    agent = QwenCodeCLIAgent()

    start = time.time()
    result = await agent.process(content)
    end = time.time()

    print(f"Total time: {end - start:.2f}s")
    # Logs include detailed timing for each step

asyncio.run(performance_analysis())

Reading logs

Log file structure

1. Main log (logs/qwen_debug.log): All levels, rotation 10 MB, retention 7 days, ZIP compression
2. Debug-only log (logs/qwen_debug_debug.log): DEBUG only, rotation 50 MB, retention 3 days, ZIP compression

Entry format

2024-10-03 15:30:45.123 | DEBUG    | qwen_code_cli_agent:_execute_qwen_cli:270 | [QwenCodeCLIAgent._execute_qwen_cli] Executing qwen-code CLI...
│                         │           │                      │                    │
│                         │           │                      │                    └─ Message
│                         │           │                      └─ Line number
│                         │           └─ Function
│                         └─ Level
└─ Timestamp

Searching logs

# Find all CLI executions
grep "CLI EXECUTION TRACE" logs/qwen_debug.log

# Find errors
grep "ERROR" logs/qwen_debug.log

# Find timing
grep "Process completed in" logs/qwen_debug.log

# Only prompts
grep -A 20 "FULL PROMPT" logs/qwen_debug_debug.log

# Only results
grep -A 50 "FULL STDOUT" logs/qwen_debug_debug.log

Security

Masking sensitive data

API keys and tokens are masked in logs:

# Environment variables:
OPENAI_API_KEY=sk-1234567890abcdef1234567890abcdef

# In logs:
OPENAI_API_KEY=sk-12345...

Protected variables

Masked variables include those containing:

• KEY
• TOKEN
• SECRET
• PASSWORD

Performance

Overhead

Level	Overhead	Log size	Recommendation
DEBUG	~5-10%	Large	Development
INFO	~1-2%	Medium	Production
WARNING	<1%	Small	Production
ERROR	<0.5%	Minimal	Production

Recommendations

• Development/Testing: DEBUG + enable_debug_trace
• Staging: INFO or DEBUG as needed; disable debug trace unless required
• Production: INFO/WARNING; enable DEBUG temporarily for incident analysis

Troubleshooting

Problem: No logs created

Solution: Ensure log directory exists and call setup_logging with DEBUG level.

Problem: Too many logs

Solution: Reduce log level to INFO or WARNING; disable debug trace.

Problem: No detailed CLI trace

Solution: Use DEBUG level and enable_debug_trace=True.

Problem: Logs too large

Solution: Adjust rotation/retention settings for log files.