Feed aggregator

Article URL: https://www.bloomberg.com/news/articles/2026-02-09/china-urges-banks-to-limit-holdings-of-us-treasuries-citing-market-volatility

Comments URL: https://news.ycombinator.com/item?id=46942467

Points: 1

# Comments: 0

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

Article URL: https://blog.coutinho.io/how-ai-tooling-changed-the-economics-of-running-a-production-grade-homelab

Comments URL: https://news.ycombinator.com/item?id=46942447

Points: 1

# Comments: 0

Article URL: https://www.wsj.com/tech/personal-tech/chinese-ev-test-drive-xiaomi-su7-c3e59282

Comments URL: https://news.ycombinator.com/item?id=46942445

Points: 1

# Comments: 0

Article URL: https://www.npr.org/2026/02/07/nx-s1-5702494/cia-world-factbook-dead

Comments URL: https://news.ycombinator.com/item?id=46942438

Points: 1

# Comments: 0

Article URL: https://eclecticlight.co/2026/02/06/apple-silicon-macs-have-2-types-of-thunderbolt-ports/

Comments URL: https://news.ycombinator.com/item?id=46942431

Points: 1

# Comments: 0

Article URL: https://www.youtube.com/watch?v=YGo5L3yzXhA

Comments URL: https://news.ycombinator.com/item?id=46942371

Points: 2

# Comments: 0

Article URL: https://matthiasott.com/articles/webspace-invaders

Comments URL: https://news.ycombinator.com/item?id=46942370

Points: 3

# Comments: 0

Article URL: https://www.genesisdb.io/blog/posts/2026-02-08/event-sourcing-database

Comments URL: https://news.ycombinator.com/item?id=46942367

Points: 4

# Comments: 0

Here are hints and the answers for the NYT Connections: Sports Edition puzzle for Feb. 9, No. 504.&

Article URL: https://www.timesofisrael.com/iran-was-notified-ahead-of-witkoff-kushner-visit-to-us-aircraft-carrier-report/

Comments URL: https://news.ycombinator.com/item?id=46942345

Points: 1

# Comments: 0

Article URL: https://www.gridbugs.org/alice-the-caml/

Comments URL: https://news.ycombinator.com/item?id=46942344

Points: 1

# Comments: 0

Article URL: https://www.iranintl.com/en/202602035280

Comments URL: https://news.ycombinator.com/item?id=46942331

Points: 1

# Comments: 0

Here are the answers for The New York Times Mini Crossword for Feb. 9.&

Article URL: https://discuss.ai.google.dev/t/gemini-3-flash-preview-inconsistent-thought-signature-generation-in-parallel-function-calls-causes-400-errors-and-potential-silent-data-loss/118936

Comments URL: https://news.ycombinator.com/item?id=46942308

Points: 1

# Comments: 0

Article URL: https://blog.zarfhome.com/2026/02/to-fight-a-troll

Comments URL: https://news.ycombinator.com/item?id=46942297

Points: 1

# Comments: 0

Article URL: https://www.seattletimes.com/seattle-news/health/cola-holy-grail-great-taste-no-calories-no-artificial-sweetener/

Comments URL: https://news.ycombinator.com/item?id=46942262

Points: 1

# Comments: 0

Article URL: https://chizkidd.github.io//2026/01/22/neural-net-optimizers/

Comments URL: https://news.ycombinator.com/item?id=46942251

Points: 2

# Comments: 0

Article URL: https://github.com/huggingface/transformers/pull/43830

Comments URL: https://news.ycombinator.com/item?id=46942237

Points: 1

# Comments: 0

The idea is simple: pipe your command through logifai, and Claude Code can search your logs directly.

npm run dev 2>&1 | logifai Hit an error → just ask Claude "what went wrong?" It captures everything into structured NDJSON, has a Web UI with live streaming, auto-redacts secrets, and includes a Claude Code Skill so Claude can search your logs without you ever copy-pasting terminal output.

It also has a Log Reference feature — select lines in the Web UI, click "Copy Ref", paste a compact logifai:// URI into Claude instead of hundreds of log lines. Keeps your context window clean.

It solves what I found to be the biggest pain point: the manual copy-paste shuttle between terminal and Claude Code.

Feedback and contributions welcome!

https://github.com/tomoyaf/logifai

Comments URL: https://news.ycombinator.com/item?id=46942233

Points: 1

# Comments: 0