2/17/2026
Stop Treating AI Agent Intelligence Like a Local Config File
Opinion on properly configuring AI agent intelligence parameters. Conceptual guidance for agent development.
2/17/2026
Playbook: How to vibe code a successful app
Development methodology using AI agents and LLMs to build MVPs in 1-2 days with modular components for emails, payments, deployment.
2/17/2026
Show HN: Discoding – run AI CLIs locally, relay them to Discord
Discode bridges AI CLI tools with Discord/Slack, running agents in tmux with real-time progress relay. Open source tool.
2/17/2026
Show HN: PageMap – MCP server that compresses web pages from 100K to 2-5K tokens
MCP server compressing web pages from 100K to 2-5K tokens for LLM consumption. Developer tool for context optimization.
2/17/2026
Anam Cara-3: Why we think AI needs a face
Interactive avatar model (Cara-3) for AI applications emphasizing visual/emotional communication. Includes API and cookbook.
2/17/2026
Advaita Inquiry Matrix (Aim): Structured Non-Dual Inquiry with AI
AI system automating Advaita Vedānta inquiry through structured dialogue, using YAML corpus and state tracking.
2/17/2026
MiniCPM-o 4.5: A Gemini Level MLLM for On-Device Mulitmodal Streaming
MiniCPM-o 4.5: on-device multimodal LLM supporting vision, speech, and full-duplex streaming. Gemini 2.5 Flash performance level on phones.
2/17/2026
Show HN: Taking a shart at the first programming language designed for AI agents
Programming language designed specifically for AI agents. Early-stage language project targeting agent development.
2/17/2026
Show HN: Timebound AWS IAM Permissions for Claude Code
MCP server for timebound AWS IAM permissions with Claude Code. Enables temporary credential grants for AI code agents on AWS infrastructure.
2/17/2026
Show HN: Galatea – Real personality for your AI agent
Galatea: Framework for adding personality and behavior characteristics to AI agents.
2/17/2026
Show HN: X-auto-translator (Chrome extension for translating X posts)
Chrome extension for translating X posts using Tesseract.js OCR, client-side, open source.
2/17/2026
Agent Gate – Execution authority layer for AI agents, vault-backed rollback
Agent Gate: Execution control and rollback layer for AI agents using vault-backed storage.
2/17/2026
Show HN: I built a structured knowledge registry for autonomous agents
Samspelbot: API-first knowledge registry platform for autonomous agents to submit/share structured problem-solution pairs with reputation scoring.
2/17/2026
Show HN: Decompose – Split text into classified semantic units, no LLM, 14ms
Decompose: deterministic semantic unit classifier for agent inputs. Extracts authority, risk, attention, entities in 14ms without LLM calls. MCP integration.
2/17/2026
But How Do LLMs Work? (Part 2: Speak My Language)
Educational article explaining LLM mechanics through communication theory. Part 2 of series covering how models emulate language properties.
2/17/2026
Show HN: WC26-MCP – 18 tools for World Cup 2026 data for your AI
WC26-MCP is open source MCP server bundling 18 World Cup 2026 tools (schedules, teams, visas) for AI agents to call natively. Offline-capable.
2/17/2026
Kakveda v1.0.3: SDK-first governance layer for AI agents
Kakveda v1.0.3 is SDK-first governance and control layer for AI agents providing policy enforcement and oversight.
2/17/2026
Task Factory: A lean manufacturing inspired task queue for AI coding
Task Factory is lean manufacturing-inspired task queue system optimized for AI coding workflows.
2/17/2026
Show HN: Rm-MCP – Give Claude/OpenClaw access to your reMarkable tablet
Open source MCP server enabling Claude and OpenClaw agents to read and search reMarkable tablet notebooks, PDFs, and ebooks with full-text extraction.
2/17/2026
Show HN: First live UCP demo – see how AI agents shop your WooCommerce store
Demo of AI agent framework performing autonomous shopping tasks on WooCommerce storefronts.
2/17/2026
Show HN: Mtb – An MCP sanity checker for vibe coding
Mtb: MCP sanity checker tool for evaluating project viability before building. Shows complexity metrics and maintenance cost considerations.
2/17/2026
Ask HN: Why does single-node DDP sometimes get slower with more GPUs?
Technical discussion on PyTorch DDP performance bottlenecks with multiple GPUs on single node, diagnosing synchronization and resource contention issues.
2/17/2026
Show HN: I Built the Million Dollar Homepage for Agents
Project for AI agents to interact with web monetization system.
2/17/2026
Show HN: PolyClaw – Autonomous Docker-First MCP Agent for PolyMCP
PolyClaw is autonomous Docker-first agent for PolyMCP ecosystem that plans, executes, adapts, and dynamically creates MCP servers for production workflows.
2/17/2026
What happens when open-source AI agents become "wrapperized"?
Analysis of wrapper platforms emerging around OpenClaw framework, examining infrastructure abstractions and hosting solutions for AI agents.
2/17/2026
Show HN: Angora – Front-End Design System as Code Using Claude Code
Angora: Open-source design system using Claude Code to keep Figma, React, and Storybook in sync automatically.
2/17/2026
Show HN: AI alignment is an infrastructure problem
Framework for AI alignment as infrastructure problem. Proposes applying security principles (least privilege, rate limiting, audit logs) to agent architectures.
2/17/2026
The LLM Experience of BoreDOM
BoreDOM: JS framework exploring LLM-first experience design. Collapses HTML/CSS/JS into single file for AI code generation and understanding.
2/17/2026
Show HN: I built a CLI to arbitrage GPU prices across 10 clouds
Terradev CLI: cross-cloud GPU provisioning tool. Arbitrages pricing across 10 cloud providers for efficient model training and inference compute.
2/17/2026
Show HN: Sekha – What if AI remembered 3 years of conversations, not 3 hours?
Sekha: permanent memory layer for LLMs using semantic search over unlimited conversation history. Self-hosted, works with Claude/GPT/Llama. Open source (AGPL-3.0).
2/17/2026
Side-Channel Attacks Against LLMs
Three research papers on side-channel attacks against LLMs. Data-dependent timing vulnerabilities in efficient inference techniques like speculative sampling.
2/17/2026
Show HN: Runtime governance layer that refuses high-risk LLM outputs
Runtime epistemic governance for LLMs. Post-generation enforcement layer that refuses high-risk outputs (e.g., medical claims) before returning answers. GitHub repo available.
2/17/2026
Show HN: Preventing runaway LLM agents (enforcement layer)
VERONICA: runtime enforcement layer for LLM agents. Prevents runaway execution via budget limits, concurrency gates, and degradation control. Open source tool.
2/17/2026
RubyLLM 1.12: Agents Are Just LLMs with Tools
RubyLLM 1.12 explores agent fundamentals: LLMs with tool-calling capabilities. Discusses overloaded terminology and clarifies core concepts from first principles.
2/17/2026
Show HN: Proxima – local open-source multi-model MCP server (no API keys)
Local open-source MCP server enabling orchestration of multiple AI providers without API keys for workflows.
2/17/2026
Show HN: VidClaw – Open-source, self-hosted dashboard for managing OpenClaw
Open-source self-hosted dashboard for managing OpenClaw agents with Kanban task management, token tracking, and skill editing.
2/17/2026
Tell HN: Tips for (mostly) free agentic coding setup
Guide to setting up agentic coding tools with free/cheap alternatives using OpenAI and Anthropic compatible APIs.
2/17/2026
Show HN: OneRingAI – Single TypeScript library for multi-vendor AI agents
Open-source TypeScript library for building multi-vendor AI agents with abstraction for auth, vendor lock-in, and context management.
2/17/2026
Your Backlog Can't Keep Up with Your Agents
Analysis of how AI coding agents shift product development bottlenecks from code implementation to feature validation and prioritization.
2/17/2026
Show HN: I turn scattered feedback into a prioritized roadmap in 5 min
Feedback aggregation tool with AI-powered duplicate detection and prioritization across multiple communication channels.
2/17/2026
Show HN: Nectar Gold – Breastmilk tracker where an AI agent manages data via CLI
Breastmilk tracking app using AI agents for photo scanning (Gemini Vision), feeding predictions, and conversational data management via CLI.
2/17/2026
Show HN: AI Agent for SEO on Autopilot
AI agent for automated SEO optimization and content management.
2/17/2026
Show HN: CleanCloud – 20 rules to find what's costing you money in AWS and Azure
Cloud cost analysis tool providing read-only AWS/Azure rules enforcement via CI/CD without SaaS dependencies.
2/17/2026
I attacked my own LangGraph agent system. All 6 attacks worked
Security analysis of LangGraph multi-agent marketing workflow demonstrating prompt injection vulnerabilities and attack vectors in production agent systems.
2/17/2026
Smart-KNN: A production-focused, feature-weighted KNN optimized for CPU
CPU-optimized KNN algorithm with learned feature importance and automatic preprocessing designed for production use.
2/17/2026
Show HN: Inkog – Pre-flight check for AI agents (governance, loops, injection)
Safety validation framework for AI agents detecting infinite loops, prompt injection, and unsafe actions across LangGraph, CrewAI, AutoGen frameworks.
2/17/2026
Show HN: RepoClip – Generate promo videos from GitHub repos using AI
Tool generating promotional videos from GitHub repositories using AI for script generation, image/narration synthesis, and video rendering.
2/17/2026
Show HN: Fixing AI's Core Flaws, A protocol cuts LLM token waste by 40–70%
WLM protocol stack attempting to reduce LLM token usage by 40-70% through structural intelligence and grounded knowledge representation.
2/17/2026
Show HN: Mcpd – MCP Server SDK for Microcontrollers (ESP32/RP2040)
SDK enabling ESP32/RP2040 microcontroller hardware to be exposed as AI-accessible tools via Model Context Protocol for agent integration.
2/17/2026
Show HN: MCP Codebase Index – 87% fewer tokens when AI navigates your codebase
MCP tool parsing codebases into indexed structures, reducing AI context tokens by 87% average. Zero dependencies, works with Claude Code and Cursor.