Kula: Self-contained Linux server monitoring with zero dependencies

What Kula does
Kula collects system metrics every second by reading directly from /proc and /sys, stores them in a built-in tiered ring-buffer storage engine, and serves them through a real-time Web UI dashboard and a terminal TUI.
Metrics collected
- CPU: Total usage (user, system, iowait, irq, softirq, steal) + core count
- Load: 1/5/15 min averages, running & total tasks
- Memory: Total, free, available, used, buffers, cached, shmem
- Swap: Total, free, used
- Network: Per-interface throughput (Mbps), packets/s, errors, drops; TCP errors/s, resets/s, established connections; socket counts
- Disks: Per-device I/O (read/write bytes/s, reads/s, writes/s IOPS); filesystem usage
- System: Uptime, entropy, clock sync, hostname, logged-in user count
- Processes: Running, sleeping, blocked, zombie counts
- Self: Kula's own CPU%, RSS memory, open file descriptors
Storage engine
Data is persisted in pre-allocated ring-buffer files per tier with fixed maximum sizes. When a file fills up, new data overwrites the oldest entries.
- Tier 1: Raw 1-second samples (default 250 MB)
- Tier 2: 1-minute aggregates: averaged CPU & network, last-value gauges (default 150 MB)
- Tier 3: 5-minute aggregates, same logic (default 50 MB)
Dashboard features
The HTTP server exposes a REST API and WebSocket endpoint for live streaming. Authentication is optional - when enabled, it uses Argon2id hashing with salt and session cookies. The frontend is a single-page application embedded in the binary with features including:
- Interactive zoom with drag-select (auto-pauses live stream)
- Focus mode to show only selected graphs
- Grid / stacked list layout toggle
- Alert system for clock sync, entropy issues, overload
Installation
Example installation methods for amd64 (x86_64) GNU/Linux. Check Releases for ARM and RISC-V packages.
Standalone
wget https://github.com/c0m4r/kula/releases/download/0.7.1/kula-0.7.1-amd64.tar.gz
echo " 6baff6bee9f9bbf56adc6e264e7ff9e1dfa763e7bab76a21dbc1e7d4be0397f4 kula-0.7.1-amd64.tar.gz " | sha256sum -c || rm kula-0.7.1-amd64.tar.gz
tar -xvf kula-0.7.1-amd64.tar.gz
cd kula
./kulaDebian/Ubuntu
wget https://github.com/c0m4r/kula/releases/download/0.7.1/kula_0.7.1_amd64.deb
echo " bbcd6ee65441c85f5bc835c40a1afaabc78b78d976c25e535c051b29ad514185 kula_0.7.1_amd64.deb " | sha256sum -c || rm kula_0.7.1_amd64.deb
sudo dpkg -i kula_0.7.1_amd64.deb
systemctl status kulaBuild from Source
git clone https://github.com/c0m4r/kula.git
cd kula
bash addons/build.shUsage
Quick Start
# 1. Copy and edit config (optional)
cp config.example.yaml config.yaml
# 2. Start the server
./kula serve
# Dashboard at http://127.0.0.1:8080
# 3. Or use the terminal UI
./kula tuiAuthentication (Optional)
# Generate password hash
./kula hash-password
# Add the output to config.yaml under web.authThe project is currently at version 0.7.1 and the author notes it still has some rough edges and needs to mature, but is already being used on multiple servers in production.
📖 Read the full source: HN LLM Tools
👀 See Also

W2A — an open protocol for agent sensors: giving local agents real-time perception
W2A (World2Agent) is an open protocol standardizing the perception layer for AI agents — self-hostable, TS SDK, Apache 2.0. It lets agents receive real-time signals from sensors without one-off scripts.

Argus: A VS Code Extension to Debug Claude Code Session Costs and Behavior
A developer built Argus, a VS Code extension that parses Claude Code JSONL transcripts into a real-time timeline with per-step token/cost breakdown, cache hit ratio, and flagging of retry loops, duplicate reads, and context pressure.

Stanford Researchers Release OpenJarvis: A Local-First Framework for On-Device AI Agents
Stanford researchers have released OpenJarvis, a local-first framework for building on-device personal AI agents with tools, memory, and learning capabilities. The project includes GitHub repository and website links for developers to explore.

CLI Tools with AI Agent Compatibility: Skills Directory Approach
A Reddit user shares a method for making CLI tools work with AI coding agents like Claude Code by creating SKILL.md files that teach agents installation, authentication, and usage. The approach addresses common issues like interactive prompts, JSON output, and authentication methods.