Skip to main content

Quick Start

Get started with OWL Watch in 5 minutes.

Step 1: Create a Project Profile​

Before monitoring logs, create a profile so OWL Watch understands your project structure:

owl-watch study ./my-project

Output:

Profiling project: ./my-project
Detected: maven project with spring-boot framework
Source dirs: src/main/java, src/main/resources
Config dirs: src/main/resources
Test dirs: src/test

Profile saved: ~/.owl-watch/profiles/my-project.json

Check existing profiles:

owl-watch study --status

Step 2: Start Monitoring​

Watch a log file with your project profile:

owl-watch ./my-project/logs/app.log --project my-project

Output:

Log file: ./my-project/logs/app.log
Project: my-project (spring-boot)
Model: llama3.2
Dashboard: http://localhost:8080

Waiting for errors...

Step 3: Open the Dashboard​

Open http://localhost:8080 in your browser.

Stats Overview​

The top of the dashboard shows error statistics:

Dashboard Stats

  • Total Errors: All detected errors
  • Pending: Errors not yet investigated
  • Resolved: Errors marked as resolved
  • With Fix: Errors that have a suggested fix

Error Timeline​

The timeline graph shows error activity over time:

Error Timeline

  • Toggle between Stream, Area, and Bar views
  • Hover to see counts at specific times
  • Filter by Pending, Resolved, or With Fix

Error List​

Below the timeline, you'll see the list of errors:

Error List

Each error shows:

  • ID: Unique identifier
  • Type: Error type (e.g., NullPointerException)
  • Status: PENDING, INVESTIGATING, or INVESTIGATED
  • Timestamp: When the error occurred

Use the filters to search and filter errors:

  • Search: Search by error text, ID, or summary
  • Status: All, Pending, Investigated, Resolved
  • Type: Filter by error type
  • Time: All Time, Last Hour, Last 24 Hours, Last 7 Days

Step 4: Investigate an Error​

Click on an error to expand it:

Error Expanded

You'll see:

  • The full error message and stack trace
  • A text field to add context (optional but helpful)

Add any relevant context, then click Investigate.

Step 5: View the Investigation​

While the LLM analyzes the error, you'll see "INVESTIGATING..." status:

Investigating

When complete, you'll see the full investigation:

Investigation Complete

The investigation includes:

Investigation Steps​

Shows what the LLM did to analyze the error:

  • thinking: LLM's analysis reasoning
  • find_files: Files the LLM searched for
  • read_file: Files the LLM read
  • search_code: Code searches performed

Fix Suggestion​

The recommended fix with:

  • Action: MODIFY, CREATE, or DELETE
  • File: The file to change
  • Explanation: Why this fixes the issue
  • Code: The actual code to apply

Actions​

  • Copy Fix: Copy the fix code to clipboard
  • OWL Prompt: View the prompt used for investigation
  • Reinvestigate: Run a new investigation with different approach
  • Mark Resolved: Mark the error as resolved

Feedback​

Rate the investigation accuracy:

  • Accurate: The fix was correct
  • Partially: The fix was partially correct
  • Inaccurate: The fix was wrong

CLI Options​

OptionDescription
--project NAMEUse a specific project profile
--port PORTDashboard port (default: 8080)
--model MODELOllama model (default: llama3.2)
--no-serverRun without dashboard (CLI only)
-v, --verboseEnable debug output

Example Workflows​

Spring Boot Application​

# Profile the project
owl-watch study ./spring-app

# Monitor the log
owl-watch ./spring-app/logs/spring.log --project spring-app

Python Django Application​

# Profile the project
owl-watch study ./django-app

# Monitor the log
owl-watch ./django-app/logs/django.log --project django-app

Using a Different Model​

# Use mistral instead of llama3.2
owl-watch app.log --project myproject --model mistral

Custom Port​

owl-watch app.log --port 9000

Keyboard Shortcuts​

In the dashboard:

KeyAction
LSwitch to Live tab
HSwitch to History tab
RRefresh data
/Focus search box

Next Steps​