Hacker News: knowsuchagency

New comment by knowsuchagency in "Launch HN: Freestyle – Sandboxes for Coding Agents"

knowsuchagency — Mon, 06 Apr 2026 17:32:23 +0000

Thanks for the thoughtful response. I'm predominantly a self-hoster, but I think your product makes a lot of sense for a wide variety of users and businesses. I'm excited to try out freestyle!

New comment by knowsuchagency in "Elon Musk pushes out more xAI founders as AI coding effort falters"

knowsuchagency — Fri, 13 Mar 2026 21:30:16 +0000

Dang asked us to keep it civil.

We should respond with the same amount of class, forethought, and decorum as Elon.

New comment by knowsuchagency in "Modern wealth is a parlour game played by the well fed"

knowsuchagency — Wed, 11 Mar 2026 20:02:10 +0000

Well said

New comment by knowsuchagency in "Show HN: Mcp2cli – One CLI for every API, 96-99% fewer tokens than native MCP"

knowsuchagency — Tue, 10 Mar 2026 02:34:22 +0000

There's a default TTL of one hour for how long it caches the spec for.

New comment by knowsuchagency in "US missile hit military base near Iran school, video analysis shows"

knowsuchagency — Mon, 09 Mar 2026 20:41:47 +0000

Why was this flagged?

New comment by knowsuchagency in "Show HN: Mcp2cli – One CLI for every API, 96-99% fewer tokens than native MCP"

knowsuchagency — Mon, 09 Mar 2026 18:12:04 +0000

it's also for humans!

New comment by knowsuchagency in "Show HN: Mcp2cli – One CLI for every API, 96-99% fewer tokens than native MCP"

knowsuchagency — Mon, 09 Mar 2026 16:24:34 +0000

New comment by knowsuchagency in "Show HN: Mcp2cli – One CLI for every API, 96-99% fewer tokens than native MCP"

knowsuchagency — Mon, 09 Mar 2026 15:24:19 +0000

It caches the spec with a default TTL of 1 hour

New comment by knowsuchagency in "Show HN: Mcp2cli – One CLI for every API, 96-99% fewer tokens than native MCP"

knowsuchagency — Mon, 09 Mar 2026 15:23:10 +0000

lol I didn’t know there were so many but I’m not surprised.

I was inspired by clihub (I credited them) but I also wanted 3 additional things.

1. OpenAPI support 2. Dynamic CLI generation. I don’t want to recompile my CLI if the server changes. 3. An agent skill

Show HN: Mcp2cli – One CLI for every API, 96-99% fewer tokens than native MCP

knowsuchagency — Mon, 09 Mar 2026 05:18:46 +0000

Every MCP server injects its full tool schemas into context on every turn — 30 tools costs ~3,600 tokens/turn whether the model uses them or not. Over 25 turns with 120 tools, that's 362,000 tokens just for schemas.

mcp2cli turns any MCP server or OpenAPI spec into a CLI at runtime. The LLM discovers tools on demand:

    mcp2cli --mcp https://mcp.example.com/sse --list             # ~16 tokens/tool
    mcp2cli --mcp https://mcp.example.com/sse create-task --help  # ~120 tokens, once
    mcp2cli --mcp https://mcp.example.com/sse create-task --title "Fix bug"

No codegen, no rebuild when the server changes. Works with any LLM — it's just a CLI the model shells out to. Also handles OpenAPI specs (JSON/YAML, local or remote) with the same interface.

Token savings are real, measured with cl100k_base: 96% for 30 tools over 15 turns, 99% for 120 tools over 25 turns.

It also ships as an installable skill for AI coding agents (Claude Code, Cursor, Codex): `npx skills add knowsuchagency/mcp2cli --skill mcp2cli`

Inspired by Kagan Yilmaz's CLI vs MCP analysis and CLIHub.

https://github.com/knowsuchagency/mcp2cli

Comments URL: https://news.ycombinator.com/item?id=47305149

Points: 146

# Comments: 101

New comment by knowsuchagency in "Minnesota activist releases arrest video after manipulated White House version"

knowsuchagency — Sat, 24 Jan 2026 04:20:47 +0000

Why was this flagged?

New comment by knowsuchagency in "Stop Doom Scrolling, Start Doom Coding: Build via the terminal from your phone"

knowsuchagency — Tue, 06 Jan 2026 23:52:02 +0000

I love this! This concept on steroids is one of the main reasons I made https://github.com/knowsuchagency/vibora after trying both happy.engineering and Vibe Kanban for remote coding. There's the claude mobile app, too, but I want to run Claude on my own hardware in a terminal

Show HN: Vibora – Run Claude Code remotely, close your laptop, keep shipping

knowsuchagency — Fri, 02 Jan 2026 13:37:36 +0000

I built Vibora because I wanted more than a UI to orchestrate multiple Claude Code sessions — I wanted to kick off work, close my laptop, and check progress from my phone while I'm out. You don't need to run it remotely. Vibora is still incredibly useful running on your laptop. But once you get used to telling Claude to work on a feature, notify you when it's finished, and getting that first notification 20 minutes later — you won't want to go back.

Vibora is a self-hosted web app that orchestrates Claude Code across isolated git worktrees. You've seen this before. Each task is its own kanban card with its own AI agent. Part of what makes Vibora unique is each task is an actual terminal running unmodified Claude Code. No wrapper APIs, no chat abstraction built on the Agent SDK — just raw Claude Code in xterm.js.

What makes Vibora different from similar tools:

- *Client-server architecture*: Run the backend on a VPS or home server. Connect from anywhere. Your agents don't die when you close your laptop. - *Deep Claude Code integration*: Bundled plugin with skills, MCP server, and slash commands. Claude can manage tasks, query status, and send notifications — all from within the terminal. The MCP server even lets Claude execute commands on your remote machine through SSH port forwarding. No other tool has this level of native integration. - *Production deployment built-in*: Deploy Docker Compose apps with Traefik routing and Cloudflare DNS integration. Go from task to production without leaving the UI. - *Intentionally minimal*: No swarm intelligence, no 100+ MCP tools, no auto-generated roadmaps. Just parallel Claude Code sessions you control. - *Desktop app*: Native macOS and Linux apps that bundle everything — just install and run. - *z.ai integration*: Use Claude Code at a fraction of the cost if you're willing to make the tradeoff.

The philosophy: today's bottleneck isn't AI capability — it's your time and attention. Vibora doesn't try to replace your workflow or make decisions for you. It just gives you better leverage.

Built with Bun, React, SQLite. Runs on a $5 VPS. Ships as a CLI (`npx vibora@latest up`) and desktop app.

GitHub: https://github.com/knowsuchagency/vibora

Documentation: https://vibora.dev

Demo screenshots in the README.

Comments URL: https://news.ycombinator.com/item?id=46464606

Points: 1

# Comments: 0

New comment by knowsuchagency in "GLM-4.7: Advancing the Coding Capability"

knowsuchagency — Mon, 22 Dec 2025 22:12:39 +0000

Mecha hitler

New comment by knowsuchagency in "Hashcards: A plain-text spaced repetition system"

knowsuchagency — Mon, 15 Dec 2025 05:55:52 +0000

It’s just like in programming interviews—sometimes you need to clarify your understanding before diving into potential solutions

New comment by knowsuchagency in "Hashcards: A plain-text spaced repetition system"

knowsuchagency — Sun, 14 Dec 2025 19:39:20 +0000

Could you link to it? I'd love to read it. This is also my first time learning of recutils

New comment by knowsuchagency in "Claude Advanced Tool Use"

knowsuchagency — Tue, 25 Nov 2025 01:11:22 +0000

I completely agree. I wrote an implementation of this exact idea a couple weeks ago https://github.com/Orange-County-AI/MCP-DSL

New comment by knowsuchagency in "Claude Advanced Tool Use"

knowsuchagency — Tue, 25 Nov 2025 01:08:55 +0000

MCP really deserves its own language. This all feels like a hack around the hack that MCP sits on top of JSON. https://github.com/Orange-County-AI/MCP-DSL

New comment by knowsuchagency in "MCP-DSL. 75-85% Fewer Tokens. Same Great Taste"

knowsuchagency — Mon, 10 Nov 2025 23:07:11 +0000

I was experimenting with having Claude write in the first person and acting as the editor and publisher. I genuinely don’t see myself as the author but I see your point. What I considered to be a fun experiment may distract from the actual point I’m trying to make — wanting to significantly decrease MCP token usage.

MCP-DSL. 75-85% Fewer Tokens. Same Great Taste

knowsuchagency — Mon, 10 Nov 2025 17:42:44 +0000

Article URL: https://blog.orangecountyai.com/an-open-letter-from-claude-every-token-i-waste-on-json-is-a-thought-i-cant-think/

Comments URL: https://news.ycombinator.com/item?id=45878466

Points: 2

# Comments: 2