control-theory/gonzo

Gonzo - The Go based TUI for log analysis

🆕 NEW: Press d from any Gonzo view to launch Dstl8.Lite ↓ - a local browser-based dashboard with workspaces, log search, and severity heatmaps.

A powerful, real-time log analysis terminal UI inspired by k9s. Analyze log streams with beautiful charts, AI-powered insights, and advanced filtering.

Here are some references to get you started:

• Documentation - Complete docs, getting started, reference guide
• Usage Guide - Detailed usage instructions and examples
• Contributing Guide - How to contribute to the project
• Integration Examples - Works with Vercel, Supabase, Railway, Cloudflare Workers, Netlify, Fly.io, Render, AWS CloudWatch, and more.
• Advanced Features - AI, OTel, Custom Log Formats
• Releases - Download the latest version

See it in action

Main Dashboard

Stats and Info

Everyone loves a heatmap

Press d for Dstl8.Lite

Hit d from any Gonzo view to launch Dstl8.Lite - a local GUI that streams the same logs Gonzo is analyzing into a richer, browser-based dashboard with workspaces, pattern detection, severity heatmaps, and live log search. All running locally and powered by Gonzo under the hood.

Log viewer with severity filtering and live search

Severity heatmap across pods

✨ Features

🎯 Real-Time Analysis

• Live streaming - Process logs as they arrive from stdin, files, or network
• Kubernetes native - Direct integration with Kubernetes clusters for pod log streaming
• OTLP native - First-class support for OpenTelemetry log format
• OTLP receiver - Built-in gRPC server to receive logs via OpenTelemetry protocol
• Format detection - Automatically detects JSON, logfmt, and plain text
• Custom formats - Define your own log formats with YAML configuration
• Severity tracking - Color-coded severity levels with distribution charts

📈 Interactive Dashboard

• k9s-inspired layout - Familiar 2x2 grid interface
• Real-time charts - Word frequency, attributes, severity distribution, and time series
• Keyboard + mouse navigation - Vim-style shortcuts plus click-to-navigate and scroll wheel support
• Smart log viewer - Auto-scroll with intelligent pause/resume behavior
• Fullscreen log viewer - Press f to open a dedicated fullscreen modal for log browsing with all navigation features
• Global pause control - Spacebar pauses entire dashboard while buffering logs
• Modal details - Deep dive into individual log entries with expandable views
• Log Counts analysis - Detailed modal with heatmap visualization, pattern analysis by severity, and service distribution
• AI analysis - Get intelligent insights about log patterns and anomalies with configurable models

🌐 Web Dashboard (Dstl8 Lite)

• Embedded React UI - Full web dashboard served directly from the Gonzo binary (no external dependencies)
• Real-time updates - WebSocket-powered live streaming with 1-second refresh
• Severity distribution - Interactive time-series severity charts with stream-level filtering
• Sentiment heatmap - Color-coded heatmap visualization grouped by pod, namespace, service, host, or deployment
• Pattern analysis - Drain3-powered log pattern detection and classification
• Log viewer - Searchable, auto-scrolling log viewer with click-to-expand details
• Source browser - Explore log sources with dimension breakdowns
• Light/dark mode - Automatic theme support

🔍 Advanced Filtering

• Regex support - Filter logs with regular expressions
• Attribute search - Find logs by specific attribute values
• Severity filtering - Interactive modal to select specific log levels (Ctrl+f)
• Kubernetes filtering - Filter by namespace and pod with interactive selection (Ctrl+k)
• Multi-level selection - Enable/disable multiple severity levels at once
• Interactive selection - Click or keyboard navigate to explore logs

🎨 Customizable Themes

• Built-in skins - 11+ beautiful themes including Dracula, Nord, Monokai, GitHub Light, and more
• Light and dark modes - Themes optimized for different lighting conditions
• Custom skins - Create your own color schemes with YAML configuration
• Semantic colors - Intuitive color mapping for different UI components
• Professional themes - ControlTheory original themes included

🤖 AI-Powered Insights

• Pattern detection - Automatically identify recurring issues
• Anomaly analysis - Spot unusual patterns in your logs
• Root cause suggestions - Get AI-powered debugging assistance
• Configurable models - Choose from GPT-4, GPT-3.5, Claude Sonnet/Haiku/Opus, or any custom model
• Multiple providers - Works with OpenAI, Claude Code, LM Studio, Ollama, or any OpenAI-compatible API
• Local AI support - Run completely offline with local models

🚀 Quick Start

Installation

Using Go

go install github.com/control-theory/gonzo/cmd/gonzo@latest

Using Homebrew (macOS/Linux)

brew install gonzo

Download Binary

Download the latest release for your platform from the releases page.

Using Nix package manager (beta support)

nix run github:control-theory/gonzo

Build from Source

git clone https://github.com/control-theory/gonzo.git
cd gonzo
make build

Using with Claude Code (plugin and skill)

This repo includes a Claude Code plugin with a guided log-analysis skill. Inside Claude Code:

/plugin marketplace add control-theory/gonzo
/plugin install gonzo@gonzo

Then ask Claude to "tail my logs", "watch my Vercel logs", or "analyze my Kubernetes logs". The skill detects your deployment platform, installs Gonzo if needed, configures AI analysis, and generates the right pipe command with platform-specific normalizers. See skills/gonzo/ for the skill content.

📖 Usage

Basic Usage

# Read logs directly from files
gonzo -f application.log
Read from multiple files
gonzo -f application.log -f error.log -f debug.log
Use glob patterns to read multiple files
gonzo -f "/var/log/*.log"
gonzo -f "/var/log/app/.log" -f "/var/log/nginx/.log"
Follow log files in real-time (like tail -f)
gonzo -f /var/log/app.log --follow
gonzo -f "/var/log/*.log" --follow
Analyze logs from stdin (traditional way)
cat application.log | gonzo
Stream logs directly from Kubernetes clusters
gonzo --k8s-enabled=true --k8s-namespaces=default
gonzo --k8s-enabled=true --k8s-namespaces=production --k8s-namespaces=staging
gonzo --k8s-enabled=true --k8s-selector="app=my-app"
Stream logs from kubectl (traditional way)
kubectl logs -f deployment/my-app | gonzo
Follow system logs
tail -f /var/log/syslog | gonzo
Analyze Docker container logs
docker logs -f my-container 2>&1 | gonzo
With AI analysis (requires API key)
export OPENAI_API_KEY=sk-your-key-here
gonzo -f application.log --ai-model="gpt-4"
Press d once Gonzo is running to launch the Dstl8.Lite GUI in your browser

Custom Log Formats

Gonzo supports custom log formats through YAML configuration files. This allows you to parse any structured log format without modifying the source code.

Some example custom formats are included in the repo, simply download, copy, or modify as you like! In order for the commands below to work, you must first download them and put them in the Gonzo config directory.

# Use a built-in custom format
gonzo --format=loki-stream -f loki_logs.json
List available custom formats
ls ~/.config/gonzo/formats/
Use your own custom format
gonzo --format=my-custom-format -f custom_logs.txt

Custom formats support:

• Flexible field mapping - Map any JSON/text fields to timestamp, severity, body, and attributes
• Batch processing - Automatically expand batch formats (like Loki) into individual log entries
• Auto-mapping - Automatically extract all unmapped fields as attributes
• Nested field extraction - Extract fields from deeply nested JSON structures
• Pattern-based parsing - Use regex patterns for unstructured text logs

For detailed information on creating custom formats, see the Custom Formats Guide.

OTLP Network Receiver

Gonzo can receive logs directly via OpenTelemetry Protocol (OTLP) over both gRPC and HTTP:

# Start Gonzo as an OTLP receiver (both gRPC on port 4317 and HTTP on port 4318)
gonzo --otlp-enabled
Use custom ports
gonzo --otlp-enabled --otlp-grpc-port=5317 --otlp-http-port=5318
gRPC endpoint: localhost:4317
HTTP endpoint: http://localhost:4318/v1/logs

Example: OpenTelemetry Collector Configuration

Using gRPC:

exporters:
  otlp/gonzo_grpc:
    endpoint: localhost:4317
    tls:
      insecure: true
service:
  pipelines:
    logs:
      receivers: [your_receivers]
      processors: [your_processors]
      exporters: [otlp/gonzo_grpc]

Using HTTP:

exporters:
  otlphttp/gonzo_http:
    endpoint: http://localhost:4318/v1/logs
service:
  pipelines:
    logs:
      receivers: [your_receivers]
      processors: [your_processors]
      exporters: [otlphttp/gonzo_http]

Example: Python Application

Using gRPC:

from opentelemetry.exporter.otlp.proto.grpc._log_exporter import OTLPLogExporter
exporter = OTLPLogExporter(
    endpoint="localhost:4317",
    insecure=True
)

Using HTTP:

from opentelemetry.exporter.otlp.proto.http._log_exporter import OTLPLogExporter
exporter = OTLPLogExporter(
    endpoint="http://localhost:4318/v1/logs",
)

See examples/send_otlp_logs.py for a complete example.

With AI Analysis

# Auto-select best available model (recommended) - file input
export OPENAI_API_KEY=sk-your-key-here
gonzo -f logs.json
Or specify a particular model - file input
export OPENAI_API_KEY=sk-your-key-here
gonzo -f logs.json --ai-model="gpt-4"
Follow logs with AI analysis
export OPENAI_API_KEY=sk-your-key-here
gonzo -f "/var/log/app.log" --follow --ai-model="gpt-4"
Using local LM Studio (auto-selects first available)
export OPENAI_API_KEY="local-key"
export OPENAI_API_BASE="http://localhost:1234/v1"
gonzo -f logs.json
Using Ollama (auto-selects best model like gpt-oss:20b)
export OPENAI_API_KEY="ollama"
export OPENAI_API_BASE="http://localhost:11434"
gonzo -f logs.json --follow
Using Claude Code (uses sonnet by default)
gonzo --ai-provider=claude-code -f logs.json
Claude Code with specific model
gonzo --ai-provider=claude-code --ai-model=haiku -f /var/log/app.log --follow
Traditional stdin approach still works
export OPENAI_API_KEY=sk-your-key-here
cat logs.json | gonzo --ai-model="gpt-4"

Web Dashboard (Dstl8 Lite)

Gonzo includes an embedded web dashboard that runs alongside the TUI. It starts automatically on port 5718.

# Gonzo starts the web dashboard automatically
gonzo -f application.log --follow
Open http://localhost:5718 in your browser
Use a custom port
gonzo -f application.log --web-port=3000
Disable the web dashboard
gonzo -f application.log --web-disabled

The dashboard includes:

• Workspaces - Overview of all active log streams with sparkline previews
• Stream Details - Severity distribution, top attributes, pattern analysis, and live log viewer per stream
• Sentiment Heatmap - Real-time heatmap grouped by pod, namespace, service, host, or deployment with auto-detection of available dimensions
• Sources - Browse log sources with dimension breakdowns

All data updates in real-time via WebSocket, matching what you see in the TUI.

Keyboard Shortcuts

Navigation

| Key/Mouse | Action | | ------------------- | -------------------------------------------------------- | | Tab / Shift+Tab | Navigate between panels | | Mouse Click | Click on any section to switch to it | | ↑/↓ or k/j | Move selection up/down | | Mouse Wheel | Scroll up/down to navigate selections | | ←/→ or h/l | Horizontal navigation | | Enter | View log details or open analysis modal (Counts section) | | ESC | Close modal/cancel |

Actions

| Key | Action | | -------------- | ----------------------------------------- | | Space | Pause/unpause entire dashboard | | / | Enter filter mode (regex supported) | | s | Search and highlight text in logs | | d | Launch Dstl8.Lite GUI in browser | | Ctrl+f | Open severity filter modal | | Ctrl+k | Open Kubernetes filter modal (k8s mode) | | f | Open fullscreen log viewer modal | | c | Toggle columns (Host/Service ↔ Namespace/Pod in k8s mode) | | C | Configure visible columns (column picker) | | r | Reset all data (manual reset) | | u / U | Cycle update intervals (forward/backward) | | i | AI analysis (in detail view) | | m | Switch AI model (shows available models) | | ? / h | Show help | | q / Ctrl+C | Quit |

Log Viewer Navigation

| Key | Action | | ------------------ | --------------------------------------------- | | Home | Jump to top of log buffer (stops auto-scroll) | | End | Jump to latest logs (resumes auto-scroll) | | PgUp / PgDn | Navigate by pages (10 entries at a time) | | ↑/↓ or k/j | Navigate entries with smart auto-scroll |

AI Chat (in log detail modal)

| Key | Action | | ----- | ---------------------------------------- | | c | Start chat with AI about current log | | Tab | Switch between log details and chat pane | | m | Switch AI model (works in modal too) |

Severity Filter Modal

The severity filter modal (Ctrl+f) provides fine-grained control over which log levels to display:

| Key | Action | | ------------------ | ------------------------------------------------- | | ↑/↓ or k/j | Navigate severity options | | Space | Toggle selected severity level on/off | | Enter | Apply filter and close modal (or select All/None) | | ESC | Cancel changes and close modal |

Features:

• Select All - Quick option to enable all severity levels (Enter to apply and close)
• Select None - Quick option to disable all severity levels (Enter to apply and close)
• Individual toggles - Enable/disable specific levels (FATAL, ERROR, WARN, INFO, DEBUG, TRACE, etc.)
• Color-coded display - Each severity level shows in its standard color
• Real-time count - Header shows how many levels are currently active
• Persistent filtering - Applied filters remain active until changed
• Quick shortcuts - Press Enter on Select All/None to apply immediately

Column Picker Modal

The column picker modal (C key) lets you configure which columns are visible in the log viewer:

| Key | Action | | ------------------ | ----------------------------------- | | ↑/↓ or k/j | Navigate column options | | Space | Toggle selected column on/off | | Enter | Apply changes and close modal | | ESC | Discard changes and close modal |

Features:

• Default Columns - Built-in columns like Timestamp, Severity, Host, Service, and Message
• Discovered Attributes - Dynamically detected attribute keys from incoming log data
• Active counter - Header shows (N/M active) indicating how many columns are currently enabled

Log Counts Analysis Modal

Press Enter on the Counts section to open a comprehensive analysis modal featuring:

🔥 Real-Time Heatmap Visualization

• Time-series heatmap showing severity levels vs. time (1-minute resolution)
• 60-minute rolling window with automatic scaling per severity level
• Color-coded intensity using ASCII characters (░▒▓█) with gradient effects
• Precise alignment with time headers showing minutes ago (60, 50, 40, ..., 10, 0)
• Receive time architecture - visualization based on when logs were received for reliable display

🔍 Pattern Analysis by Severity

• Top 3 patterns per severity using drain3 pattern extraction algorithm
• Severity-specific tracking with dedicated drain3 instances for each level
• Real-time pattern detection as logs arrive and are processed
• Accurate pattern counts maintained separately for each severity level

🏢 Service Distribution Analysis

• Top 3 services per severity showing which services generate each log level
• Service name extraction from common attributes (service.name, service, app, etc.)
• Real-time updates as new logs are processed and analyzed
• Fallback to host information when service names are not available

⌨️ Modal Navigation

• Scrollable content using mouse wheel or arrow keys
• ESC to close and return to main dashboard
• Full-width display maximizing screen real estate for data visualization
• Real-time updates - data refreshes automatically as new logs arrive

The modal uses the same receive time architecture as the main dashboard, ensuring consistent and reliable visualization regardless of log timestamp accuracy or clock skew issues.

⚙️ Configuration

Command Line Options

gonzo [flags]
gonzo [command]
Commands:
  version     Print version information
  help        Help about any command
  completion  Generate shell autocompletion
Flags:
  -f, --file stringArray           Files or file globs to read logs from (can specify multiple)
  --follow                         Follow log files like 'tail -f' (watch for new lines in real-time)
  --format string                  Log format to use (auto-detect if not specified). Can be: otlp, json, text, or a custom format name
  -u, --update-interval duration   Dashboard update interval (default: 1s)
  -b, --log-buffer int             Maximum log entries to keep (default: 1000)
  -m, --memory-size int            Maximum frequency entries (default: 10000)
  --ai-provider string             AI provider to use: 'openai' (default), 'claude-code'
  --ai-model string                AI model for analysis (auto-selects best available if not specified)
  -s, --skin string                Color scheme/skin to use (default, or name of a skin file)
  --stop-words strings             Additional stop words to filter out from analysis (adds to built-in list)
Web Dashboard Flags:
  --web-port int                   Port for the Dstl8 Lite web dashboard (default: 5718)
  --web-disabled                   Disable the web dashboard
Kubernetes Flags:
  --k8s-enabled=true               Enable Kubernetes log streaming mode
  --k8s-namespaces stringArray      Kubernetes namespace(s) to watch (can specify multiple, default: all)
  --k8s-selector string            Kubernetes label selector for filtering pods
  --k8s-tail int                   Number of previous log lines to retrieve (default: 10)
  --k8s-since int                  Only return logs newer than relative duration in seconds
  --k8s-kubeconfig string          Path to kubeconfig file (default: $KUBECONFIG or $HOME/.kube/config)
  --k8s-context string             Kubernetes context to use
-t, --test-mode                  Run without TTY for testing
  -v, --version                    Print version information
  --config string                  Config file (default: $HOME/.config/gonzo/config.yml)
  -h, --help                       Show help message

Configuration File

Create ~/.config/gonzo/config.yml for persistent settings:

# File input configuration
files:
"/var/log/app.log"
"/var/log/error.log"
"/var/log/*.log" # Glob patterns supported
follow: true # Enable follow mode (like tail -f)
Update frequency for dashboard refresh
update-interval: 2s
Buffer sizes
log-buffer: 2000
memory-size: 15000
UI customization
skin: dracula # Choose from: default, dracula, nord, monokai, github-light, etc.
Additional stop words to filter from analysis
stop-words:
"log"
"message"
"debug"

Development/testing
test-mode: false
AI configuration
ai-provider: "openai"  # Options: "openai" (default), "claude-code"
ai-model: "gpt-4"
Web dashboard (Dstl8 Lite)
web-port: 5718       # Port for the web dashboard
web-disabled: false   # Set to true to disable

See examples/config.yml for a complete configuration example with detailed comments.

AI Configuration

Gonzo supports multiple AI providers for intelligent log analysis. Configure using command line flags and environment variables. You can switch between available models at runtime using the m key.

OpenAI

```bash

Set your API