2. Show HN: Why – a deterministic curiosity graph built on probabilistic model (why.com)
One question. Three paths. Go deeper.
3. Show HN: Millennium – Steam Homebrew (steambrew.app)
An open-source low-code modding framework to create, manage and use themes/plugins for the desktop Steam Client without any low-level internal interaction or overhead.
4. Show HN: Hacker News minus the slop (hnfiltered.com)
Hacker News, unchanged, minus stories whose discussion suggests that clicking the link will be a waste of time.
6. Show HN: E3d-pilot – a repo-improving agent harness, SHA-gated merges (github.com)
Contribute to spacepacket1/e3d-pilot development by creating an account on GitHub.
7. Show HN: Agentic Task Management (github.com)
An idea file to create an agentic task management system - myaa2913/agentic-task-management
8. High-Performance Array-Backed LRU Hash Table (github.com)
A cross-platform, multi-core optimized C++20 LRU Hash Table featuring zero runtime allocations, custom TTAS spinlocks, and Windows Kernel support. - adanil-code/LRUHashTable
9. Show HN: Getting review comments out of Overleaf, and into the PDF (github.com)
Export Overleaf comments and tracked changes to Markdown + JSON, optimized for AI-agent consumption - Mangluu/overleaf-comments-export
10. Show HN: Captain, AI Travel Agent (www.moonlight.ng)
An AI travel agent on Telegram. Talk through any trip to organize plans, look up live prices, and prepare.
13. Show HN: Hexis, open-source Claude Skills management (github.com)
Git-backed control plane for AI-agent skills, tools, context, permissions, and identity. Self-hosted and MCP-native. - Bevel-Software/Hexis
15. Show HN: Genosyn – Automate running of companies with AI (genosyn.com)
Open-source, self-hosted company operating system where Members and AI Employees share context, tools, work, and review queues.
16. Why Books Don't Work: Constructivism over Transmissionism (2021) (www.masterhowtolearn.com)
17. Show HN: Control Claude Code, Codex, Pi and Gemini CLI from Telegram (github.com)
A Telegram bot that drives four local coding CLIs — Claude Code, Codex, Pi, and Gemini — from your phone. macOS daemon with per-chat per-agent sessions, a confirm gate for dangerous commands, respo...
18. Show HN: Auto-train the harness, not the LLM. cross-model, cross-benchmark gains (github.com)
Train a harness to improve its model-agnostic and task-env-agnostic capabilities with PyTorch-like APIs - workofart/harness-training
19. Compute-Optimal Is Not Cluster-Optimal (szha.ai)
Our new paper folds the systems stage into the scaling-law stage. Price every candidate architecture on what the cluster actually delivers, and the answer changes: the sparsity an MoE
A recent paper challenges traditional views on model architecture efficiency by linking performance to the specific characteristics of the training cluster. It reveals that the optimal sparsity of a model varies depending on the cluster it operates within, suggesting a need for a more tailored approach to architecture selection.
20. Show HN: Wevna – Local-first runtime dashboard for Node.js back ends (github.com)
Local-first runtime dashboard for Node.js backends. Watch HTTP, SQL, Redis and console activity live, correlated per request. - wevna/wevna
21. Show HN: Made a business license hub search tool (fernway.io)
Track every annual report and franchise tax for every LLC and corporation you own, whoever formed them and wherever registered. All 50 states and D.C.
22. Show HN: A simple list for concerts in NYC (singledigit.nyc)
SingleDigit is a plain list of upcoming live music around New York City.
24. Show HN: Karma in Menu Bar (github.com)
A native macOS menu bar app for tracking Hacker News karma. - rorz/karma-bar
25. Show HN: DST this DST that, test your containers (github.com)
Test the **** out of containers. Deterministic Simulation Testing for containerised services. - bxrne/dstest
26. Show HN: Self-bench – build SWE-bench style evals from private repos (github.com)
Build Harbor evals from a GitHub repo. Contribute to mupt-ai/self-bench development by creating an account on GitHub.
28. Show HN: SVC16 = a specified virtual computer to build compilers for (janneuendorf.github.io)
29. Show HN: Remarc – contextual feedback for coding agents via MCP (github.com)
Your feedback layer for AI collaboration. Point at anything on your Mac - text, screenshots, web elements, or voice - and your agent reads and resolves your comments over MCP. - metedata/Remarc