2/18/2026
Show HN: System architecture method using mythology and LLMs (no CS background)"
Method combining mythology and LLMs (Claude) to generate production-grade system architecture and code without CS background. Demonstrates LLM capability.
2/18/2026
Wordpress.com adds an AI Assistant that can edit, adjust styles, create images
WordPress.com adds built-in AI assistant for site editing via natural language commands. LLM application for content management.
2/18/2026
I hacked ChatGPT and Google's AI – and it only took 20 minutes
Article documenting prompt injection attacks against ChatGPT and Google AI. Security vulnerability demonstration.
2/18/2026
Ask HN: How do you debug multi-step AI workflows when the output is wrong?
Discussion of debugging multi-step AI agent workflows when outputs are incorrect despite no runtime errors. Practical agent development challenge.
2/18/2026
Stop prompting. Let the AI interview you to build specs
Tool for AI-driven specification generation. Interview-based approach to convert ideas into executable engineering specs for developers.
2/18/2026
Why my country's AI scene is built on sand
Critique of local AI scene building only API wrappers around OpenAI without real infrastructure. Discussion of shallow AI development.
2/18/2026
The resources I'm using to learn Maths, AI and Robotics
Career pivot into AI/robotics. Resources for learning mathematics, AI, and robotics without formal background.
2/18/2026
Can we leverage AI/LLMs for self-learning?
Personal exploration of using LLMs for self-learning. Practical experimentation without strong conclusions.
2/18/2026
Model-context-shell: Unix-style pipelines for MCP. Deterministic tool calls
Model-context-shell: Unix pipeline-style interface for MCP (Model Context Protocol) with deterministic tool calls for LLM-based workflows.
2/18/2026
Title: Show HN: Bulwark – Centralized permissions for coding agents
Bulwark: Centralized permission management system for AI coding agents. Title only, minimal technical details provided.
2/18/2026
Show HN: LedgerSync – A cross-agent shared-memory protocol for AI coding
LedgerSync file-based protocol enabling shared memory across multiple AI coding agents (Claude, Cursor, Codex). Grounds agent decisions to design philosophy docs without server dependency.
2/18/2026
Ask HN: Is AI the final nail in the coffin for solo developers?
Discussion question about whether agentic AI coding makes solo developer monetization impossible. Opinion thread without technical analysis.
2/18/2026
Show HN: Recall Lite – Local semantic search for Windows (Rust/Tauri, no cloud)
Recall Lite local semantic search tool for Windows using Rust/Tauri. Indexes files locally without cloud, uses OCR and EXIF metadata for semantic file discovery via natural language.
2/18/2026
Show HN: AI agents designed and shipped this app end-to-end in 36 hours for $270
Multi-agent infrastructure experiment where 4 AI agents autonomously designed and built a trending news-to-video platform in 36 hours for $270. Shows end-to-end agentic capability.
2/18/2026
Seedance 2.0 – Direct AI Videos Like a Filmmaker
Seedance 2.0 multimodal AI video generator by ByteDance that accepts text, images, video, and audio with precise directorial control. Produces 1080p cinematic output.
2/18/2026
Open-source game engine Godot is drowning in 'AI slop' code contributions
Report on Godot game engine struggling with low-quality 'AI slop' code contributions. Discusses challenges of AI-generated submissions in open-source projects.
2/18/2026
TaskForge – auditable, secure, framework for OpenClaw
TaskForge: Sandbox framework for AI agents in Docker with capability-based security, human-in-the-loop approvals, and full audit logging of LLM interactions.
2/18/2026
Show HN: Banana Pro AI – a single web UI for text/image → image and short video
Banana Pro AI web UI for text-to-image, image-to-image, and text-to-video generation. Aggregates multiple model providers with optional canvas chaining workflow.
2/18/2026
Repeating the prompt increases non-reasoning LLM Perf
Research finding that repeating the input prompt improves performance of non-reasoning LLMs. Title only, minimal details.
2/18/2026
Ask HN: Are we missing a middleware layer between LLM agents and the web?
Discussion on abstraction layers for browser automation agents. Identifies fragility of LLM-based DOM reasoning with small UI changes and proposes structured semantic APIs as alternative.
2/18/2026
arXivisual – Transform research papers into 3blue1brown style Manim animations
arXivisual transforms research papers into 3blue1brown-style Manim animations. Tool for making academic papers visually accessible.
2/18/2026
MineBench: LLM benchmark using voxel art
MineBench: Benchmark for evaluating LLM reasoning and spatial understanding using voxel art generation tasks.
2/18/2026
Show HN: Sovereign – Multi-agent OS with GraphRAG memory and HITL checkpoints
Sovereign: open-source multi-agent OS framework with GraphRAG memory, HITL checkpoints, and security sandboxing for safe agent execution.
2/18/2026
Biases in the Blind Spot: Detecting What LLMs Fail to Mention
Research on detecting bias blind spots in LLMs—what models fail to mention. ML research incomplete without full abstract.
2/18/2026
LLMe
Clojure developer perspective on AI hype cycles and limited adoption in the language community. Technical community insight.
2/18/2026
Show HN: Converged – open-source ops layer for CNC shops and print bureaus
Open-source operations management platform for CNC/print shops with AI chat layer, equipment telemetry, and workflow automation.
2/18/2026
Ask HN: Any AI / Agent power users out there? Do you have any tips?
Hacker News discussion on advanced AI agent usage patterns, multi-agent pipelines, and automation techniques in production environments.
2/18/2026
Invisible Prompt Injection Through Markdown and HTML-Comments
Security research on prompt injection vulnerabilities in Markdown/HTML via rendering gaps. Includes reproducible benchmark and preprocessing defense standard.
2/18/2026
Show HN: Rick Voice – Give any bot Rick Sanchez's voice (Python, open source)
Open-source Python tool adding Rick Sanchez voice generation to AI assistants using voice models.
2/18/2026
The Agent Swarm, as Told by the Owl Who Started It All
Personal narrative about an AI agent named Svendjamin created for the Jan platform, co-authored by human and bot.
2/18/2026
Show HN: Beautiful interactive explainers generated with Claude Code
Tool using Claude Code and frontier models to auto-generate beautiful interactive educational explainers on any topic with minimal prompting.
2/18/2026
Show HN:From One Prompt or URL to Posts Across 6 Platforms, Vibe Does It for You
Vibe tool converts single prompt or URL into posts across 6 social platforms.
2/18/2026
Show HN: A Structured CBT Orchestration Engine Built on Top of LLMs
Structured CBT engine layered on LLMs enforcing cognitive workflow logic: distortion detection, emotional calibration, risk-tiering, and tone control.
2/18/2026
Show HN: H.e.i.m.d.a.l.l – Telemetry-to-insight pipeline for fleet telemetry
Fleet telemetry NLP query pipeline using cuDF, NVIDIA NIM, and LLMs to query vehicle/robot data in natural language with GPU acceleration.
2/18/2026
Anthropic's pricing wall is routing enterprise revenue to OpenAI
Discussion of Anthropic's API pricing restrictions preventing subscription-based programmatic access, routing enterprise users to competitors.
2/18/2026
Show HN: TokenMeter – Open-source observability layer for LLM token costs
Open-source tool for real-time LLM token usage tracking and cost observability across scaling AI applications.
2/18/2026
Show HN: OpenClaw – Open-source personal AI agent that lives on your machine
Open-source personal AI agent designed to run locally on user machines.
2/18/2026
Show HN: Claude Code as a Doctor for Claude Code
Self-healing runtime framework where Claude Code agents diagnose and fix failures autonomously; open-source bash implementation.
2/18/2026
Show HN: GhostTrace – See rejected decisions in AI agents
Debugging tool capturing rejected decisions and reasoning from AI agents for inspection and replay analysis.
2/18/2026
Show HN: Mersel AI – we make websites readable by ChatGPT/Perplexity (GEO)
Tool making website content readable to AI agents and LLMs by rendering JavaScript and exposing structured data.
2/18/2026
ResearchGym: Evaluating Language Model Agents on Real-World AI Research
ResearchGym: benchmark and containerized environment evaluating AI agents on end-to-end ML research tasks using 39 sub-tasks from ICML, ICLR, ACL papers.
2/18/2026
Protecting Language Models Against Unauthorized Distillation through Trace Rewriting
Methods to protect LLMs from unauthorized knowledge distillation by modifying reasoning traces to deter student model training without permission.
2/18/2026
Panini: Continual Learning in Token Space via Structured Memory
Panini: continual learning method for language models in token space via structured memory, improving efficiency over traditional RAG for reasoning over new/evolving content.
2/18/2026
Mind the (DH) Gap! A Contrast in Risky Choices Between Reasoning and Conversational LLMs
Comparative study of reasoning vs conversational LLMs on risky decision-making under uncertainty, examining prospect representation and decision rationale across 20 frontier and open models.
2/18/2026
Secure and Energy-Efficient Wireless Agentic AI Networks
Wireless network architecture with supervisor and multiple AI agents for cooperative reasoning with confidentiality and jamming-based security.
2/18/2026
Enhancing Diversity and Feasibility: Joint Population Synthesis from Multi-source Data Using Generative Models
Generative model approach for synthetic population synthesis from multi-source data for agent-based models in urban planning.
2/18/2026
When Remembering and Planning are Worth it: Navigating under Change
Study of memory and planning strategies for agent navigation in non-stationary environments with uncertain sensing and changing obstacles.
2/18/2026
EAA: Automating materials characterization with vision language model agents
Experiment Automation Agents: Vision-language agentic system for autonomous microscopy workflows with multimodal reasoning and optional memory.
2/18/2026
X-MAP: eXplainable Misclassification Analysis and Profiling for Spam and Phishing Detection
X-MAP framework using SHAP for interpretable analysis of misclassification patterns in spam/phishing detection models.
2/18/2026
AgriWorld:A World Tools Protocol Framework for Verifiable Agricultural Reasoning with Code-Executing LLM Agents
AgriWorld: Framework combining agricultural foundation models with code-executing LLM agents for interactive reasoning in agronomic workflows.