Hacker News

AI Project Handoff Format

Hacker News - Sat, 03/07/2026 - 9:19pm
Categories: Hacker News

Show HN: DailyDefense – Daily tower defense for agents or humans

Hacker News - Sat, 03/07/2026 - 9:17pm

A simple tower defense game. Seeded randomly every day. Get the top score.

Have your agent try to get the top score.

dailydefense.ai/skill.md

Comments URL: https://news.ycombinator.com/item?id=47293696

Points: 1

# Comments: 0

Categories: Hacker News

Show HN: Too many AI SaaS launching every day so we built Arena where they fight

Hacker News - Sat, 03/07/2026 - 9:13pm

Every day a new “AI tool” launches. Most are vibecoded clones.

So we built Glad-ia-tor : an arena where AI SaaS fight for survival.

The crowd roasts them. The crowd votes. Only the strongest survive.

Comments URL: https://news.ycombinator.com/item?id=47293666

Points: 1

# Comments: 0

Categories: Hacker News

Show setup modal with confetti on coverage page when no CI data exists

Hacker News - Sat, 03/07/2026 - 9:12pm

Coverage dashboard used to show a tiny setup link when no CI data existed. Most users missed it. Now a modal pops up on page visit with a "Set Up" button that creates the CI workflow PR in one click. Archived and empty repos get a different message instead of a broken setup flow.

Comments URL: https://news.ycombinator.com/item?id=47293650

Points: 1

# Comments: 0

Categories: Hacker News

Ask HN: How to serve inference as we do with containes with cached token

Hacker News - Sat, 03/07/2026 - 9:04pm

I've been reading and experimenting with vLLM but it seems that each day there are more and more articles and AI generated long form about each part of the stack. I have a few GPUs and work for a private education group. I want to run models internally and distribute access to a research team. I don't want to have one (or more) GPU per user neither train models. CUrrently I am doing well with a local Qwen on my own single server but I can't wrap my head around on which part to tackle - right now I am looking to KV caches and building over vLLM but I wanted something simple and secure to not leak data from one session to another.

Comments URL: https://news.ycombinator.com/item?id=47293590

Points: 1

# Comments: 0

Categories: Hacker News

The HArc Stack – A Web Stack Built on Raku

Hacker News - Sat, 03/07/2026 - 8:39pm

Article URL: https://harcstack.org

Comments URL: https://news.ycombinator.com/item?id=47293436

Points: 1

# Comments: 0

Categories: Hacker News

Show HN: Apc-CLI – sync AI memory across Claude Code, Cursor, Copilot

Hacker News - Sat, 03/07/2026 - 8:37pm

I use 4-5 AI coding tools depending on the task. The thing that kept frustrating me: every tool stores its context in a different place, in a different format. Switch from Claude Code to Cursor mid-project and you start from scratch. Set up a new machine and you rebuild everything by hand.

apc-cli fixes this with three commands:

apc collect # extract skills, MCP servers, and memory from every installed tool apc status # see what's in sync and what isn't apc sync # push everything to all your tools It handles: Claude Code · Cursor · Gemini CLI · GitHub Copilot · Windsurf · OpenClaw

No cloud account. Everything runs locally. API keys go to the OS keychain, not a config file. MIT licensed.

PyPI: https://pypi.org/project/apc-cli/

Happy to hear what's missing — especially if you use tools I haven't added support for yet.

Comments URL: https://news.ycombinator.com/item?id=47293427

Points: 1

# Comments: 0

Categories: Hacker News

Pages