Hacker News: ziyasal

New comment by ziyasal in "The Token Compression Illusion: Why I'm Skeptical of RTK"

ziyasal — Thu, 18 Jun 2026 20:40:02 +0000

> Mainstream CLIs and developer tools can easily ship a native --compact or --json-stream flag tailored for LLM consumption.

Until they do, they won't soon , rtk, caveman, ponytail and many others are just trying to address every growing costs (for 2K org, its around 2.5M, for now), so these are trade-offs we are all know and adjusting, but unlike the author claims we know the trade-off well and forking these tools, benchmarking, verifying the output quality matches our needs and so on to make it work for us, so no blindly.

For solo devs, yes, they might not really need it, self hosting another model to save would be better option. But for orgs thats a spicy part.

Yes, its good that we see these articles are shedding some light but like we do with these tools, lets also consume these articles with a grain of salt.

New comment by ziyasal in "AI IDE Plugin: Did you get it?"

ziyasal — Mon, 01 Jun 2026 14:59:37 +0000

Quietly fixes messy prompts — typos, dropped words, thumb-typed, half-awake — so Models gets it the first time.

Supports: Claude Code, Cursor, OpenCode

AI IDE Plugin: Did you get it?

ziyasal — Mon, 01 Jun 2026 14:59:37 +0000

Article URL: https://github.com/bugthesystem/dygit

Comments URL: https://news.ycombinator.com/item?id=48357700

Points: 2

# Comments: 1

Show HN: Agentjail – Self Hosted Freestyle.sh

ziyasal — Sun, 19 Apr 2026 21:39:00 +0000

Minimal Linux sandbox for running untrusted code. Built for AI agents, build systems, and any scenario where you need to execute code you didn't write.

Comments URL: https://news.ycombinator.com/item?id=47827862

Points: 2

# Comments: 0

New comment by ziyasal in "Agentjail: Minimal Linux sandbox for running untrusted code/apps/agents"

ziyasal — Sun, 12 Apr 2026 18:17:40 +0000

Minimal Linux sandbox for running untrusted code. Built for AI agents, build systems, and any scenario where you need to execute code you didn't write.

Agentjail: Minimal Linux sandbox for running untrusted code/apps/agents

ziyasal — Sun, 12 Apr 2026 18:17:40 +0000

Article URL: https://github.com/bugthesystem/agentjail

Comments URL: https://news.ycombinator.com/item?id=47742665

Points: 1

# Comments: 1

New comment by ziyasal in "Ask HN: What are you working on? (January 2026)"

ziyasal — Mon, 12 Jan 2026 12:29:52 +0000

I've been building a platform that lets you describe a multiplayer game idea in natural language and generates full-stack code for it—frontend (PixiJS, Phaser, Babylon, or Three.js) plus backend (KaosEngine real-time multiplayer framework). You can iterate via chat, preview live, sync to GitHub, and deploy.

Stack overview:

Networking — Rust-powered WebSocket + RUDP with fast ring buffers — https://github.com/bugthesystem/Kaos

Game Server Framework — Rust-powered Matchmaking, Lua runtime with hot-reload, leaderboards, storage, chat, rooms, social features (friends, etc.)

SDKs — Rust, TypeScript/JavaScript, Unity, Godot, Defold

Game Studio — AI-assisted builder (similar to Lovable.dev), Monaco-based Lua/HTML IDE, live preview with hot-reload for both backend and frontend, one-click deploy

Early results are promising. Would love feedback from folks who've worked on similar problems or have thoughts on the approach.

Planning to spin up a Discord and public playground in a week or two for anyone interested in early access.

– Sia

New comment by ziyasal in "Ask HN: How can we make our AI/MCP Security product reach audiences?"

ziyasal — Sat, 20 Dec 2025 20:08:36 +0000

Hey thanks for the reply, standalone would be an option noted, also we have audit logs, and Splunk integration support but we will improve "auditability", thanks for the suggestions.

Ask HN: How can we make our AI/MCP Security product reach audiences?

ziyasal — Fri, 19 Dec 2025 20:57:35 +0000

We've been building with AI tools and noticed there wasn't a good way to manage MCP servers across a team or see what's actually flowing to LLM providers. Who's running what? Which tools are approved? What data is going where or whats shared on AI websites? So we built CyberCage ().

# What it does:

- MCP Management — Auto or manual discovery of MCP servers, with approval workflows. Manage allowed MCP servers org-wide (down to individual tools). Secure MCP catalog (integrates with GitHub's MCP Catalog).

- Operations — Manage allowed AI applications org-wide. Full audit logs (Splunk integration available). Notifications via Slack, Teams, Webex, webhooks.

- Network inspection for configured AI domains (but not limited to) to have PII detection, private data exfiltration, de-anonymization and masking

# Works with: AI IDEs: Claude Code, Cursor, VS Code, Windsurf, Antigravity. Low-code platforms: n8n (native integration).

# In private beta:

On-device network agent for configured AI domains. Content inspection for PII and sensitive data. Packet metadata anomaly analysis.

# Coming soon:

BYOLLM (bring your own models for inspection). Browser extensions.

See it in action:

We built this for visibility and control over AI tooling without slowing teams down.

P.S. We're planning to open source CyberSmol v1.0 — a small model fine-tuned for AI threat detection — once it's ready.

Happy to answer questions ♥

Comments URL: https://news.ycombinator.com/item?id=46330830

Points: 1

# Comments: 2

New comment by ziyasal in "Show HN: CyberCage – Security platform for AI tools and MCP servers"

ziyasal — Sat, 13 Dec 2025 10:37:59 +0000

Thanks for the comment, the inspection coverage going to be available all network interactions with AI domains (in private beta) not only MCPs in a few weeks.

New comment by ziyasal in "Show HN: CyberCage – Security platform for AI tools and MCP servers"

ziyasal — Thu, 11 Dec 2025 18:54:07 +0000

Here is the full list of supported AI Assistants and IDEs: Claude Desktop, Claude Code, VS Code, Cursor, Windsurf, Zed, Kiro, Codex CLI, Antigravity, IntelliJ IDEA, and n8n (<https://www.npmjs.com/package/@cybercage/n8n-nodes-mcp-tool>).

More: https://docs.cybercage.io/

Show HN: CyberCage – Security platform for AI tools and MCP servers

ziyasal — Thu, 11 Dec 2025 18:29:15 +0000

So we built CyberCage (<https://cybercage.io>).

What it does:

MCP Management — Auto or manual discovery of MCP servers, with approval workflows. Manage allowed MCP servers org-wide (down to individual tools). Secure MCP catalog (integrates with GitHub's MCP Catalog).

Operations — Manage allowed AI applications org-wide. Full audit logs (Splunk integration available). Notifications via Slack, Teams, Webex, webhooks.

Works with:

AI IDEs: Claude Code, Cursor, VS Code, Windsurf, Antigravity. Low-code platforms: n8n (native integration).

In private beta:

On-device network agent for configured AI domains. Content inspection for PII and sensitive data. Packet metadata anomaly analysis.

Coming soon:

BYOLLM (bring your own models for inspection). Browser extensions.

See it in action: <https://youtu.be/Zy7XhkQkUlk>

We built this for visibility and control over AI tooling without slowing teams down.

P.S. We're planning to open source CyberSmol v1.0 — a small model fine-tuned for AI threat detection — once it's ready.

Happy to answer questions ♥

Comments URL: https://news.ycombinator.com/item?id=46235072

Points: 6

# Comments: 5

New comment by ziyasal in "Ask HN: What are you working on? (July 2025)"

ziyasal — Sun, 27 Jul 2025 20:01:05 +0000

I'm working on this, improving in every iteration. (Documentation needs to be updated as well)

https://github.com/bugthesystem/Flux

Flux is a high-performance message transport library for Rust that implements patterns inspired by LMAX Disruptor and Aeron. It provides lock-free inter-process communication (IPC), UDP transport, and reliable UDP with optimized memory management for applications with low latency requirements.

New comment by ziyasal in "Flux is a high-performance, zero-copy message transport library for Rust"

ziyasal — Sun, 20 Jul 2025 17:36:17 +0000

Ahaha, yeah it fails to align thats true, but still not that bad on documenting

(BTW documentation needs to be updated, I already put a note for this concern in the Readme ..)

New comment by ziyasal in "Flux is a high-performance, zero-copy message transport library for Rust"

ziyasal — Sat, 19 Jul 2025 15:44:29 +0000

Flux is a high-performance, zero-copy message transport library for Rust, designed for ultra-low-latency applications requiring maximum throughput. It is built on patterns & practices inspired by LMAX Disruptor and Aeron, with modern Rust optimizations.

Flux is a high-performance, zero-copy message transport library for Rust

ziyasal — Sat, 19 Jul 2025 15:44:29 +0000

Article URL: https://github.com/bugthesystem/Flux

Comments URL: https://news.ycombinator.com/item?id=44616472

Points: 21

# Comments: 3

New comment by ziyasal in "PhantomGPU: GPU performance emulator to benchmark ML models on virtual GPUs"

ziyasal — Mon, 07 Jul 2025 10:49:44 +0000

Please bear in mind: Its still in development to increase accuracy and cover more models but could be useful for research purposes.

New comment by ziyasal in "PhantomGPU: GPU performance emulator to benchmark ML models on virtual GPUs"

ziyasal — Mon, 07 Jul 2025 10:48:24 +0000

PhantomGPU is an -experimental- GPU performance emulator that lets you benchmark ML models on virtual GPUs with validated accuracy. Compare performance across different GPUs and estimate costs without access to physical hardware.

PhantomGPU: GPU performance emulator to benchmark ML models on virtual GPUs

ziyasal — Mon, 07 Jul 2025 10:48:24 +0000

Article URL: https://github.com/bugthesystem/PhantomGPU

Comments URL: https://news.ycombinator.com/item?id=44488834

Points: 3

# Comments: 2

Show HN: Fast thread-safe in-memory cache that supports big number of entries Go