Hacker News

Subscribe to Hacker News feed
Hacker News RSS
Updated: 30 min 28 sec ago

Show HN: Tiles – yet another Emacs package for note-taking

Mon, 02/09/2026 - 2:53am

In TILES, each note (or tile, if you will) is a single plaintext paragraph stored in its own .org file, organized through tags and bold keywords. TILES tries to keep it simple: there is no note title (to reduce friction when creating a note, and honestly a note title is not something I find very useful anyway), there are no dependencies (except for Emacs, version 27.1 at least), no links between notes, no backlinks, no graphs, and no database; every note is a paragraph in its own plaintext file.

Other notable features are: - a dashboard offering a quick glance and a preview of your notes; - note stitching (borrowed from Howm) and the possibility to get note content by using Dynamic Blocks in Org Mode (borrowed from Denote); - tags are mandatory and used for hierarchy; - keywords are optional and automatically extracted from bolded content inside the notes; - search is performed exclusively on tags and/or keywords; - subnotes (or undertiles, if you will), a kind of meta-content (or private content) inside a note, which is a paragraph prefixed with '&&' hidden everywhere except in expanded view in the dashboard and, of course, in the note editing buffer (it's not exported trough Dynamic Blocks actions, nor through stitching); - color coding in the dashboard, depending on the note's age.

Comments URL: https://news.ycombinator.com/item?id=46942698

Points: 1

# Comments: 0

Categories: Hacker News

Show HN: MCPlexor – MCP multiplexer that cuts agent context usage by 95%

Mon, 02/09/2026 - 2:17am

I built MCPlexor to solve a token waste problem I kept running into with MCP-based agents.

The Problem: MCP (Model Context Protocol) is great for giving LLMs access to external tools. But if you connect multiple servers (GitHub, Linear, Postgres, Slack), you end up with 40-50k tokens of tool definitions injected into every request – before the agent even does anything.

On a 200k context model, that's 25% gone. On smaller models, it's worse. And most runs only use 1-2 tools.

The Solution: MCPlexor sits between your agent and your MCP servers. Instead of loading all tool definitions upfront:

Agent asks for a capability ("create an issue") MCPlexor routes to the right server using semantic matching Only relevant tools get exposed Result: ~500 tokens overhead instead of ~20k.

Technical Details: - Written in Go, single binary, no runtime deps - Supports both stdio and HTTP transports - Stores credentials in OS keychain (macOS Keychain, Windows Credential Manager, Linux keyring via zalando/go-keyring) - Works with Claude Desktop, Cursor, Augment Code, or any MCP-compatible client

The routing logic can run entirely locally using Ollama. No API calls, no cost, works offline.

Business Model (for transparency): - Local/Ollama users: completely free - Cloud tier (on waitlist): we run inference on efficient small models instead of Opus/Pro, pass on savings, take a margin The bet is that routing is a narrow enough task that a fine-tuned 7B model does it as well as a frontier model. Early testing suggests this works.

CLI uploads: https://github.com/arustagi101/mcplexor Install: `curl -fsSL https://mcplexor.com/install.sh | bash` Happy to answer questions about the architecture, the routing approach, or anything else.

Comments URL: https://news.ycombinator.com/item?id=46942466

Points: 1

# Comments: 0

Categories: Hacker News

Web­Space Invaders

Mon, 02/09/2026 - 1:58am
Categories: Hacker News

Alice the Caml

Mon, 02/09/2026 - 1:55am
Categories: Hacker News

'Help is on the way': to whom?

Mon, 02/09/2026 - 1:53am
Categories: Hacker News

To Fight a Troll

Mon, 02/09/2026 - 1:46am
Categories: Hacker News

Adding Support for Qwen3.5

Mon, 02/09/2026 - 1:38am
Categories: Hacker News

Pages