Nov 16 Maker News - Browser Online Office and Agent Tools
Translated by AIView original
1. [New service] An online space to work together in a browser dot office
Indiehacker Online is a service where solo founders and developers gather in a pixel-art office in the browser and each work on their own. It brings over the effect of seeing the person next to you work and staying focused, and it offers a Pomodoro timer, a to-do list, and project tools, all free. The server runs on a single Elixir Phoenix instance, and the author says it's built to send only each moving person's position, at short intervals.
Maker point If you want to add a sense of working together to a productivity tool you use alone, you can try out the approach of sending only small data like position on a regular basis.
Source: GeekNews Show GN
2. [New service] Multiple coding agent usage in one dashboard
agent-otel-kit is a tool that shows usage, costs, and tool calls from Claude Code, Codex, and Gemini CLI together on one screen. Running docker compose up once locally brings up the collector, storage, and a Grafana dashboard together, and a small Go CLI hooks in each agent's settings. The creator says it isn't meant to replace the existing /usage screen, but is a tool for individuals and teams who use several agents.
Maker point If you switch between three or more agents and want to know costs by period, spin it up locally with docker compose and check last week's spending first.
Source: GeekNews Show GN
3. [New service] The San Francisco Bay Area as a pixel map
This project takes Google Photorealistic 3D Tiles data and turns roughly 1,300 square miles of the San Francisco Bay Area into a single pixel map. You can go looking for landmarks like Apple Park, the Palace of Fine Arts, and UC Berkeley on the screen. A technical write-up is posted alongside it, so it's easy to follow how it was made.
Maker point Reducing real city 3D data to low-resolution pixels is a good starting point if you want to build a small game map with your own neighborhood as the backdrop.
Source: Hacker News Show HN
4. [Open source] NVIDIA's evaluation framework for measuring agent skills
NVIDIA released SkillEvaluator, a multi-stage framework for evaluating AI agent skills. It supports quality gates, duplicate detection to find skills that overlap in meaning, synthetic evaluation dataset generation, and live agent evaluation. It focuses on measuring how much a skill actually changes an agent's behavior.
Maker point If you're not sure whether the agent skill you built actually changes behavior, run a few cases through a synthetic dataset and record the results as numbers, and the direction for your next fix becomes clear.
Source: GitHub Trending
5. [Open source] A coding agent skill that trims your tokens by shortening how it talks
caveman is an open-source skill and proxy that makes coding agents talk briefly, and it claims to cut tokens significantly. It presents benchmark numbers showing it costs less than other agents on the same tasks. Recently it also released a new version called Caveman Code.
Maker point If you're a solo developer feeling the weight of your API bill, put the proxy in front and compare token usage before and after on the same tasks.
Source: GitHub Trending
Do you use any tools that help you stay focused when you work alone?
Comments 0
No comments yet
Be the first to comment.