Showcases on Lukas Manera

tracepack

Mon, 06 Apr 2026 00:00:00 +0000

tracepack is a small Go CLI that quickly scans codebases for patterns and saves the results as Markdown, designed to stay simple, fast, and flexible through YAML profiles.

Stack

Why I built it

When you inherit a large, old, or unfamiliar codebase, the first problem is usually not deep semantic analysis. It is getting a fast, reusable overview of what is there: size, structure, hotspots, and recurring patterns worth reviewing first.

Tradeoffs

The tool is intentionally lightweight and pattern-driven. That keeps it flexible and easy to adapt with YAML profiles, but it also means it is not a replacement for deeper static analysis, framework-aware tooling, or manual review.

Notes

The most useful output is often a compact footprint plus a saved Markdown bundle of searches and command output. That makes triage, migration planning, legacy reviews, and sharing findings with others much easier.

It supports two modes:

hyprorbit

Sat, 18 Oct 2025 00:00:00 +0000

hyprorbit is a workspace orchestration tool for Hyprland designed to preserve muscle memory while switching between projects or operating contexts.

Stack

Why I built it

Switching between projects usually meant sacrificing my window arrangement or rebuilding it. I wanted the same predictable, project-based layout available in multiple contexts at once. Hyprland did not make that easy on its own.

Tradeoffs

It adds complexity. Moving windows across orbits needs extra keybindings, and selecting windows or configuring window rules with multiple monitors gets a bit messy and takes more setup than plain Hyprland. It also would probably be better as a Hyprland plugin than as an external layer.

Notes

A daemon/client split was worth it for low-latency. It fits my workflow but I rarely have more than one orbit running at a time tbh so much of it could be achieved with hyprland on its own.

Built as a Go daemon and CLI, hyprorbit uses persistent IPC, YAML-driven configuration, and direct Hyprland integration to orchestrate workspace focus, window placement, and orbit switching with minimal latency.

TaskVanguard

Fri, 20 Jun 2025 00:00:00 +0000

Stack

Screenshot

Why I built it

Task managers are great at storing tasks but dont help much with starting them. TaskVanguard tries addressing the friction between knowing what to do and actually doing it — lowering the activation energy through LLM-powered task reframing, impact tagging, and context-aware spot suggestions.

Tradeoffs

It works best when pulling a specific task forward from urgency, tags, and current context. Restructuring large backlogs or generating projects dynamically is weaker, because the model does not have enough context to understand the full shape of the work.

Notes

The best part therefore is still the ‘spot’ workflow: lowering friction around the next actionable task.

Status

Released. Available via go install and AUR (yay -S taskvanguard).

MATA

Wed, 02 Apr 2025 00:00:00 +0000

MATA is a lightweight monitoring dashboard for PHP-heavy environments where full observability stacks are impractical.

Stack

Screenshot

Why I built it

I wanted monitoring that still works on shared hosting, constrained VPSes, and legacy PHP environments where you cannot install agents or extra services. Existing stacks were usually heavier than the problem called for.

Tradeoffs

It is intentionally not a full observability platform. The design stays small and deployable through read-only PHP nodes and pull-based collection, which fits small server fleets well but is not meant for large-scale or Kubernetes-style monitoring.

Notes

The useful part is combining uptime checks, PHP logs, server metrics, Composer inventory, and alerts in one place. OPA handles alarm decisions cleanly, Taskfile became part of the day-to-day workflow for setup, deploys, and user management, and optional Docker support keeps deployment flexible.

Status

Released. Demo available at demo.mata.sh.