Feature Benchmark and Roadmap

This document records vminfo's feature tradeoffs after comparing similar open-source projects. It is a planning reference for issues and PR review, not a promise list.

Product position

vminfo should stay within these boundaries:

• Single binary, low dependency, low configuration — install and use immediately without a daemon, database, or central server.
• Local, real-time diagnostics first — TUI, CLI, JSON, Web/API all focus on seeing the current machine quickly.
• Scriptable and embeddable — keep the JSON schema and Go public API stable for CI, automation, and other Go tools.
• Lightweight web, not a monitoring platform — borrow the instant readability of Netdata, but not the full cloud, multi-node, long-retention stack.
• Clear platform semantics — summary / watch stay cross-platform; ps / kill remain Linux-only.

Reference projects

Project	Representative strengths	What vminfo can borrow
btop	High-quality TUI, resource graphs, process filtering, tree view, signal actions, themes	Improve TUI visuals and process interaction
bottom / btm	Cross-platform terminal monitoring, widget focus, layout config, history windows	Add focus/basic modes and better narrow/wide layouts
Glances	TUI, web, REST API, JSON/CSV/stdout, client/server, MCP	Strengthen script output and API docs
procs	Modern ps replacement, search, tree, watch, IO, Docker info	Make vminfo ps a practical troubleshooting entrypoint
htop	Mature process operations, sorting, filtering, kill/renice, command display	Align TUI process page with familiar shortcuts
Netdata	Zero-config realtime web, automatic discovery, clear summaries	Borrow the “one glance” web summary without platform complexity
gopsutil / psutil	Stable cross-platform collection APIs	Keep the high-level Go API stable and well documented

Recommended roadmap

P0: Highest value

1. Enhance vminfo ps

   • vminfo ps <keyword>: filter by process name, user, PID, or command line.
   • vminfo ps --filter <keyword>: explicit filter for scripts.
   • vminfo ps --tree: tree view.
   • vminfo ps --watch: continuous refresh with --count and --interval.
   • vminfo ps --limit 20: limit rows after sort/filter.
   • vminfo ps --sort cpu|mem|pid|name: keep the current sort set.

2. Add process detail views

   • Show PID, PPID, user, command line, CPU, RSS, and start time in the TUI.
   • Keep the Web process table and API filterable and sortable.

3. Add a web problem summary

   • Show top CPU, memory, disk, and network warning signals.
   • Keep it read-only, stateless, and lightweight.

P1: Automation and integration

4. Field selection and tabular output

   • Add user-selectable field subsets for summary and watch.
   • Keep field names aligned with JSON.

5. Short history windows

   • Keep recent CPU, memory, network, and disk I/O samples in memory.
   • Avoid databases and default disk writes.

6. Go library examples

   • Add more import examples for single snapshots, continuous sampling, custom JSON, and embedded TUI.

P2: Useful, but not core

7. MCP server

   • Expose snapshot and process tools for AI assistants.

8. Container identification

   • Show Docker/containerd identifiers in ps, TUI, and Web views.

9. Lightweight health scoring

   • Emit a transparent score and warnings set.

Not recommended

• Long-term data storage or databases
• Cloud accounts or multi-node control planes
• Heavy plugin systems
• Default Prometheus/Influx exporters
• A large frontend app for the dashboard

Delivery rules

• If CLI behavior, JSON output, Web API, or exported Go APIs change, update README, docs, and tests together.
• Keep ps / kill Linux-only.
• Keep web remote access guarded by token, cookie, CORS, and WebSocket origin rules.