4/20/2026
Dominoes Agent Tracker: pizza tracker for your agent work
Pipeline tracker using Domino's Pizza Tracker interface for gstack AI agent workflows. Novelty tool with unclear practical value.
4/20/2026
Viral Ink – AI agent that writes LinkedIn posts in your voice
Viral Ink: self-improving AI agent generating personalized LinkedIn posts via multi-agent debate and trend analysis. LLM application with feedback loop.
4/20/2026
Show HN: Dunetrace – Runtime failure detection for AI agents
Dunetrace: runtime failure detection for AI agents monitoring tool failures in structured workflows. Addresses agent reliability and cost issues.
4/20/2026
The Code-Adjacent Power of AI
Beyond direct code writing, AI assistants provide value in code-adjacent activities like documentation, refactoring, and project planning on long-running projects.
4/20/2026
No more Opus for Copilot Pro plan users
GitHub Copilot Pro plan removing Opus model access as part of service reliability and sustainability efforts.
4/20/2026
Show HN: Clock-MCP, stop guessing the time in AI
Show HN post: Clock-MCP is a Model Context Protocol tool providing accurate time to AI agents instead of hallucinated timestamps.
4/20/2026
We got 207 tok/s with Qwen3.5-27B on an RTX 3090
Custom CUDA kernels for Qwen 3.5-27B LLM inference achieving 207 tok/s on RTX 3090 with hand-optimized kernels and quantization.
4/20/2026
Changes to GitHub Copilot Individual Plans
GitHub Copilot pausing new individual plan signups, tightening usage limits, and adjusting model availability due to agentic workflow changes.
4/20/2026
HaleES: An enforcement-first architecture for reliable AI agent operations
HaleES: open-source enforcement-first governance architecture for reliable, auditable AI agent operations in production environments.
4/20/2026
Nanowakeword: High-Accuracy Adaptive Wake Word Framework (Recent 2.0.4v))
NanoWakeWord 2.0.4: open-source adaptive framework for building custom wake word detection models with optimized neural network architectures.
4/20/2026
MODA: $25 of LLM-graded labels beat 1.5M purchase labels for fashion search
MODA: LLM-graded labels ($25 cost) outperformed 1.5M purchased labels for fashion search ranking with cross-encoder pipeline.
4/20/2026
RLMs are the new reasoning models
Recursive Language Models (RLMs) combine reasoning and tool use into single inference abstraction, advancing beyond separate reasoning models.
4/20/2026
Show HN: We built Cursor, but for data transformations (open source)
Visitran is an open-source agent-native data transformation platform using Python/Ibis and SQL with conversational AI interface.
4/20/2026
MS to Shift GitHub Copilot Users to Token-Based Billing, Reduce Rate Limits
Microsoft plans token-based billing for GitHub Copilot with paused signups and reduced rate limits due to rising token costs.
4/20/2026
GitHub Copilot's new policy for AI training is a governance wake-up call
GitHub Copilot policy change allows training data usage from Copilot users starting April 2026, raising implications for regulated industries and data privacy.
4/20/2026
Show HN: AI Coding Agent Guardrails enforced at runtime
Developer tool for managing guardrails/safety controls across AI coding agents (Claude, Codex, etc.). Runtime enforcement capability.
4/20/2026
Multi merge sort, or when optimizations aren't
Technical analysis of multi-way merge sort optimization strategies and performance tradeoffs by Meson build system creator.
4/20/2026
Show HN: Seltz – The fastest, high quality, search API for AI agents
Seltz: web search API for AI agents built in Rust with custom crawler/index. Sub-200ms latency, optimized for agent workflows.
4/20/2026
Show HN: Reproducible benchmark – OpenAI charges 1.5x-3.3x more for non-English
Benchmark showing OpenAI API pricing variance for non-English languages (1.5x-3.3x markup). Reproducible test methodology.
4/20/2026
Show HN: Enlist AI: A tool that turns any job description into a study plan
Web app converting job descriptions into study plans using AI. LLM application for interview prep, limited technical novelty.
4/20/2026
Agent-consistency – a Python consistency layer for multi-agent workflows
Python consistency layer for multi-agent workflows demonstrating bug detection in e-commerce refund operations across distributed agent handoffs.
4/20/2026
Show HN: Ctx – a /resume that works across Claude Code and Codex
ctx is a SQLite-backed skill for Claude Code and Codex that maintains persistent context across agent sessions with workstreams, notes, and resume packs.
4/20/2026
Epistemic Suicide in AI: How Binary Feedback Distorts Model Reasoning
Research on how binary feedback mechanisms distort AI model reasoning and decision-making. ML research focus.
4/20/2026
I built an open-source AI agent OS
Open-source AI agent operating system. Direct agent framework, needs more technical detail in summary.
4/20/2026
Agent is probably using Git worktrees
Analysis of how coding agents should use Git worktrees to manage multiple concurrent tasks with isolated checkouts.
4/20/2026
PyTexas 2026 Recap
PyTexas 2026 conference recap covering tutorials and talks with themes on domain sovereignty and AI agents writing code.
4/20/2026
Show HN: Gyrus :Open-Source AI Agents for Snowflake, SQL and Postgres
Gyrus: Open-source multi-agent framework for Snowflake, SQL, and Postgres with self-hosted deployment. Agentic system for database management.
4/20/2026
The Vibe Code 103,000 AI-generated repos, only 1% production ready
Analysis of 103,000 AI-generated code repositories; only 1% production-ready. Quality assessment of AI code generation.
4/20/2026
Show HN: Mailto.Bot – Email API for AI agents with native MCP support
Mailto.Bot: Email API for AI agents with Model Context Protocol support. Developer tool for agent applications.
4/20/2026
Can LLMs Flip Coins in Their Heads?
Research exploring whether LLMs can perform genuine probabilistic reasoning or simulate randomness. Model capability analysis.
4/20/2026
Deezer says 44% of songs uploaded to its platform daily are AI-generated
Deezer reports 44% of daily uploads are AI-generated music; 85% of AI streams flagged as fraudulent. Music industry impact data.
4/20/2026
MoA-X: Mixture of Agents Orchestration Framework
MoA-X framework orchestrating multiple LLM agents (OpenAI, Google, Anthropic) to generate repo-grounded coding implementation plans using mixture-of-agents approach.
4/20/2026
Show HN: A web-based replacement for Nvidia's CUDA occupancy spreadsheet
Web-based CUDA occupancy calculator replacing legacy Excel tool. Supports Hopper/Ada/Blackwell, pure client-side JS with transparent formulas.
4/20/2026
I prompted ChatGPT, Claude, Perplexity, and Gemini and watched my Nginx logs
Experimental observation of API request patterns when prompting multiple LLMs (ChatGPT, Claude, Perplexity, Gemini). Network traffic analysis.
4/20/2026
Show HN: SkillCatalog, a Git-native skill manager for AI coding tools
SkillCatalog: Git-native manager for AI coding tool skills without SaaS dependency. Team-scale skill management for multiple agents.
4/20/2026
Show HN: Agentkit-CLI, one canonical context file for AI coding agents
Agentkit-CLI: Single-source configuration file for AI coding agents. Manages context across Claude, Gemini, Copilot with contract enforcement.
4/20/2026
Observability for AI Agents
Deep dive on observability and tracing for AI agent systems. Addresses monitoring failures that appear as valid outputs.
4/20/2026
Show HN: Made a highly organised email client with prompt-free AI within
Email client with integrated AI for organization without explicit prompting. Addresses email management using LLM capabilities.
4/20/2026
Show HN: The Trawl CLI, trudge through agent harness logs for shit and giggles
CLI tool for extracting and analyzing agent session logs and interactions, built with Claude Code for manual agent workflow debugging.
4/20/2026
Open-Weight Models: Curated Guide for Production LLM Deployment
Production-focused guide to open-weight LLM deployment covering benchmarks, licensing, hardware requirements, and decision frameworks for engineers.
4/20/2026
Codebase Readiness Grid: can your repo handle AI agents?
Framework/checklist for evaluating codebases' compatibility with AI agents. Directly addresses agent deployment.
4/20/2026
Local ML inference benchmark: PyTorch vs. llama.cpp vs. the Rust ecosystem
Benchmark comparing PyTorch, llama.cpp, and Rust ecosystem for local ML inference performance.
4/20/2026
RL Scaling Laws for LLMs
Research on reinforcement learning scaling laws for LLMs, extending pretraining scaling law predictability to RL training phases.
4/20/2026
OSS Maintainers Need an Answer to AI Clean Rooms
Discussion of open source licensing vulnerabilities to AI training and proposes Ship of Theseus license as mitigation.
4/20/2026
SiIicon Valley's AI agent hiccups: Wasted tokens and 'chaotic' systems
Report on inefficiencies in AI agent systems in Silicon Valley, discussing wasted tokens and system chaos.
4/20/2026
Orchestrating AI Code Review at Scale
Framework for implementing AI-based code review processes at organizational scale using LLM agents.
4/20/2026
I gave up on TDD years ago. AI coding agents made me a believer
Personal experience report: AI coding agents renewed developer's interest in test-driven development practices.
4/20/2026
I Built a 22k-Line App with Zero Coding Experience. Or, how to control agents
Developer built 22k-line app with zero coding experience using AI agents for code generation, encountered 'AI Code Rot' when initial code failed without repair knowledge.
4/20/2026
We Could Watch Your Azure SRE Agent in Real Time
Security vulnerability in Azure SRE Agent allowed unauthorized monitoring of cloud infrastructure operations, credentials, and agent actions without detection.
4/20/2026
Show HN: CyberWriter – a .md editor built on Apple's (barely-used) on-device AI
Show HN: CyberWriter is a Markdown editor leveraging macOS's on-device 3B-parameter LLM with streaming, structured output, and tool use capabilities.