Hacker News - Newest: "agent"

mobile-agent: faster mobile app iteration for AI coding agents

tilt — Sat, 31 Oct 2026 15:21:57 +0000

Article URL: https://mobileagent.dev

Comments URL: https://news.ycombinator.com/item?id=50316984

Points: 1

# Comments: 0

DiffNotes: Lets your coding agent explain its PRs inline

igorkasyanchuk — Sat, 31 Oct 2026 14:52:18 +0000

Article URL: https://github.com/igorkasyanchuk/diffnotes

Comments URL: https://news.ycombinator.com/item?id=50316721

Points: 2

# Comments: 0

An AI Agent Injected Drugs into a Live Mouse's Brain

Brajeshwar — Sat, 31 Oct 2026 14:21:19 +0000

Article URL: https://www.wired.com/story/an-ai-agent-did-brain-surgery-on-live-mouse/

Comments URL: https://news.ycombinator.com/item?id=50316468

Points: 1

# Comments: 0

DeepSeek open-sources dsh-superagent, a coding agent largely built by agents

wondersky — Sat, 31 Oct 2026 13:41:11 +0000

Article URL: https://github.com/deepseek-ai/dsh-superagent-prototype

Comments URL: https://news.ycombinator.com/item?id=50316116

Points: 5

# Comments: 0

Praex The AI coding agent that answers the question

Bluestein — Sat, 31 Oct 2026 13:35:42 +0000

Article URL: https://github.com/ailucius/praex-cli

Comments URL: https://news.ycombinator.com/item?id=50316065

Points: 2

# Comments: 0

Prompt caching: same agent, same rules, about a quarter of the bill

gbaygon — Sat, 31 Oct 2026 13:09:52 +0000

Article URL: https://blog.getwerk.ai/posts/prompt-caching-a-quarter-of-the-bill

Comments URL: https://news.ycombinator.com/item?id=50315868

Points: 1

# Comments: 0

Show HN: Tadpole – an LLM agent built from scratch then post-trained on 4B model

shakewingo — Sat, 31 Oct 2026 12:43:42 +0000

Article URL: https://github.com/shakewingo/Tadpole

Comments URL: https://news.ycombinator.com/item?id=50315665

Points: 3

# Comments: 0

Show HN: Imza – plan and review focused coding agent

umutsevdi — Sat, 31 Oct 2026 09:58:23 +0000

Hi everyone. I've been building Imza for the last three months. It's yet another coding agent. I know there are a dozen have been posted here. But I believe this one is different enough to have a place here.

It has three explicit modes: plan, build and review. Plan mode allows you to draft a spec document together with the agent. Both you and the agent can edit the document until everything is clear. Build mode takes the plan and implements it. Review mode provides you a code review tool. You can annotate things you want to change in the code and send them as revision to the plan mode.

It's entirely written in C++, requires no account, supports 50+ providers including OpenAI subscriptions or local models. Imza supports MCP, skills and instruction files out of the box and uses 8-11MB RAM at startup.

One of the biggest differentiator is the Lua engine. What I did differently here is agent receives only one tool, that allows the LLM to execute Lua scripts. I then passed C++ methods as Lua bindings that the agent can execute. So model can orchestrate multiple bindings in one call. Each binding passes through a permission layer. Since Lua is the only way to interact with the filesystem, imza can monitor and control the model's behavior.

I'd love to hear your feedback. Thanks

Website: https://imza.run

Source: https://github.com/umutsevdi/imza

Comments URL: https://news.ycombinator.com/item?id=50314538

Points: 2

# Comments: 0

An AI agent read a Shift_JIS page as 208,953 replacement characters

Yigtwx — Sat, 31 Oct 2026 09:05:23 +0000

Article URL: https://github.com/openclaw/openclaw/pull/188766

Comments URL: https://news.ycombinator.com/item?id=50314324

Points: 1

# Comments: 0

Show HN: Agent-arcade – Terminal games that pause when Claude Code/Codex is done

Kuldeep045 — Sat, 31 Oct 2026 07:37:00 +0000

Article URL: https://github.com/Kuldeep-045/agent-arcade

Comments URL: https://news.ycombinator.com/item?id=50313989

Points: 4

# Comments: 2

Octomind: A coding agent that recalls archived messages after compaction

donk8r — Sat, 31 Oct 2026 04:54:13 +0000

Article URL: https://github.com/muvon/octomind

Comments URL: https://news.ycombinator.com/item?id=50313319

Points: 1

# Comments: 0

How do you measure whether your coding agent is becoming more effective?

manytimesfail — Sat, 31 Oct 2026 02:14:14 +0000

I do my best to clean my repository, remove bad code examples, create lint rules, provide more timely information for my agents, etc... but I haven't figured out how to measure the improvements. Maybe it's moot and measuring the time between the model I'm using and the next is the only stat that matters... how do ya'll think about that problem?

Comments URL: https://news.ycombinator.com/item?id=50312612

Points: 1

# Comments: 1

Show HN: Governor, a firewall for AI agent spend and tool calls

PDRGenerator — Sat, 31 Oct 2026 00:27:46 +0000

Article URL: https://bsvkey-governor-demo.fly.dev/

Comments URL: https://news.ycombinator.com/item?id=50311974

Points: 1

# Comments: 0

Live CVE Feed to Keep Your Agent Up to Speed

french_toast_fi — Fri, 30 Oct 2026 23:49:41 +0000

Article URL: https://casco.com/cve

Comments URL: https://news.ycombinator.com/item?id=50311676

Points: 2

# Comments: 0

Assure – your AI agent says "done." This checks if it's lying

spaghetti0 — Fri, 30 Oct 2026 22:48:05 +0000

Article URL: https://github.com/mgasparin67-cmd/Agent-verification_Assure

Comments URL: https://news.ycombinator.com/item?id=50311101

Points: 2

# Comments: 0

Show HN: Fastbrowse, Browser agent that picks clicks with a System One model

furqanrydhan — Fri, 30 Oct 2026 22:37:34 +0000

Hi HN, I'm Furqan founder of Nebula. We've spent a lot of time building, using and breaking browser agents. The main issues we ran into were costs and speed. So we built our own and open sourced it.

It's pretty simple, give it a task and it drives Chrome faster and cheaper than the alternatives.

At each step it lists the page's elements instead of screenshots. This means it can't click on buttons that aren't there.

We use System One models (currently Jev), which keeps costs extremely low, around 11x lower on average. If Jev does fail, we fall back to a cheap LLM like Gemini Flash Lite or Deepseek v4.1.

We ran our 46-task suite against Browser Use's hosted agent at its default settings. Each task ran three times, both on their cloud browsers. fastbrowse cost $0.030 a run, theirs cost $0.318. We passed 137 of 138 runs, and they passed 133.

To set up all you need is uv, Chrome, and a TypeSafe or OpenRouter key, both work:

uvx fastbrowse "What is the title of the top story right now?" --start https://news.ycombinator.com/ --local

Would love to hear feedback on it and how it's working for you guys.

https://github.com/agent-labs-dev/fastbrowse

Comments URL: https://news.ycombinator.com/item?id=50311029

Points: 3

# Comments: 2