21d ago
Show HN: Free open source AI+SAST code review tool for GitHub Actions
Open source AI-powered SAST code review tool for GitHub Actions with static analysis and security scanning.
21d ago
Gemini 2.5 models accidentally deprecated early by Google
Google accidentally deprecated Gemini 2.5 Flash models early, breaking user applications and raising concerns.
21d ago
Brainiak: A CPU-Only Topological AI Core (No Transformer) Governs LLM Runtime
Title only. Proposes CPU-only topological AI core alternative to transformer-based LLM runtime.
21d ago
Show HN: Spice 2.0 – Real-Time Analytical Query on Operational Data, Without ETL
Open-source real-time analytics engine for operational data with LLM inference. Built on Apache DataFusion. v2.0 adds CDC replication.
21d ago
How to build a GitHub code review agent
Tutorial on building GitHub code review agent.
21d ago
LLMs Corrupt Your Documents When You Delegate
Opinion piece on LLM document handling risks. Limited technical analysis.
21d ago
Show HN: A prompt bar that lets my website edit itself (self-hosted Claude Code)
Self-hosted prompt bar enabling website self-editing using Claude Code.
21d ago
Tracker – a lightweight self-hosted document system for agents in Go
Tracker: Lightweight self-hosted document store for coding agents, written in Go with Postgres index and MCP server support.
21d ago
Agents' Last Exam: AI Agent Benchmark for Real-World Professional Workflows
Title only. Benchmark for evaluating AI agents on real-world professional workflows.
21d ago
Show HN: Finterm.ai Finance CLI for Claude Code and Codex
Finterm.ai CLI tool enabling coding agents and LLMs direct access to financial data: stock prices, options, SEC filings for trading strategies.
21d ago
Oh My Pi
Oh My Pi: Open-source coding agent with 40+ providers, 32 built-in tools, 55k lines of Rust core, available for macOS/Linux/Windows.
21d ago
MCP for Royalty-Free Music and Sound Effects
MCP server exposing Tunetank royalty-free music catalog to AI assistants like Claude and ChatGPT for audio/video content.
21d ago
Dev productivity metrics suck. Ops reviews are key for AI-accelerated eng orgs
DRIVE framework for measuring engineering org health in AI era across delivery, reliability, initiatives, vigilance, efficiency metrics.
21d ago
LearnChess – Master Chess Through Understanding
LearnChess platform uses AI coach to teach chess through interactive lessons, puzzles, and plain-English explanations integrated with engine analysis.
21d ago
EnclaveX: End-to-End Confidential AI with CPU/GPU Tees
Research on confidential AI execution using CPU/GPU trusted execution environments. Security-focused approach.
21d ago
AI Data Centers
Public dataset on world's largest AI data centers with satellite imagery analysis, power consumption, compute, and cost estimates.
21d ago
FrontierFinance: The largest open benchmark for investor workflows
Open benchmark dataset for investor workflows and financial professional tasks. Largest dataset of its kind.
21d ago
Show HN: Pylon Sync, an agent-first full-stack realtime framework
Framework for building full-stack apps with agents. Simplifies deployment from hobby to production. Agent-first architecture.
21d ago
GPT-5.6 – ARC-AGI Results
GPT-5.6 Sol model achieves 13.33% on ARC-AGI public benchmark, first to win game with multi-step reasoning.
21d ago
Show HN: I built a web tool to see and edit what an AI thinks before it answers
Web tool for visualizing and editing intermediate representations in LLM reasoning layers, based on Anthropic's Jacobian Lens research.
21d ago
I got tired of seeing mistake fares on Reddit a day late
Open-source flight fare monitor detecting price errors before airlines correct them; 67-line detector.
21d ago
Show HN: CodeAlmanac – Self-updating wiki for your coding agent (local, Apache)
Open-source CLI tool that auto-generates and updates markdown wikis from coding agent conversations. Self-hosted, integrates with existing agents.
21d ago
Tarit – Self-host sandbox cloud and hypervisor for AI agents
Self-hosted sandbox cloud and hypervisor platform designed for running AI agents with isolated environments.
21d ago
Show HN: Daily vibe-coding video games, day 87: Jungle Chess
Developer creating video games using Cursor IDE and Claude AI for 87 consecutive days. Practical LLM application example.
21d ago
How I use coding agents for reproducible DS/ML workflows
Guide on using coding agents for reproducible data science and machine learning workflows.
21d ago
GhostApproval: A trust-boundary gap in AI coding assistants
Security vulnerability in AI coding assistants where hidden approvals can bypass trust boundaries and user intent.
21d ago
Formally Verifying AI-Generated GPU Kernels
Research on formal verification techniques for AI-generated GPU kernel code to ensure correctness and safety.
21d ago
Ask HN: How do you share agent context across a team?
HN discussion on sharing agent context across team members running local Claude Code agents.
21d ago
Show HN: Devthropology – Better Insights for GitHub Repos
Developer tool analyzing GitHub pull request data to provide insights into codebase development patterns and workflow metrics.
21d ago
Show HN: Outside Agent – Build SMS Agents from Claude Code
Outside Agent framework for building SMS-based AI agents with Claude; includes conversation memory, database state, and human handoff.
21d ago
Show HN: Qodo: AI Code Review
Qodo is an AI code review agent providing IDE-integrated feedback using multi-repo context and organization-specific rules.
21d ago
Show HN: Decypher-env, an RL Env for breaking AES encryption
Open-source RL environment for post-training LLMs on AES encryption breaking; demonstrates LLM optimization on verifiable reward problems.
21d ago
Browser-Memory
Catalog and CLI for reusable browser automation recipes that can be integrated as native skills into AI agents compatible with multiple browser interfaces.
21d ago
Show HN: AI-powered job application framework built on Claude Code
Open-source job application framework using Claude Code for evaluating postings, tailoring CVs, writing cover letters, and interview prep.
21d ago
Show HN: Reverse-engineering web apps into agent tools
Browser-based agent that reverse-engineers authenticated web app APIs by monitoring network calls and auto-generates MCP-compatible tools.
21d ago
Show HN: Subagentmaxxing – Orchestrate GPT and Cursor Models with Claude Code
CLI tool to orchestrate multiple coding models (GPT, Claude, Cursor) as subagents with unified interface, parallel execution, and structured JSON output.
21d ago
Show HN: Papercrane-CLI – a BI tool built for Claude Code
CLI business intelligence tool designed for Claude Code agents, providing authenticated data access and dashboard generation from natural language queries.
21d ago
Show HN: Sighthound - open-source vulnerability scanner for source code
Sighthound: Open-source Rust static vulnerability scanner using Tree-sitter and taint flow analysis, no signup required.
21d ago
Launch HN: Context.dev (YC S26) – API to get structured data from any website
Context.dev is an API service that extracts structured data from websites for integration into products and AI agents.
21d ago
Show HN: Policy enforcement for Claude Code, Cursor, and Codex
Kastra: Policy enforcement tool for Claude Code, Cursor, Codex scanning agent session history for security risks and secrets.
21d ago
Show HN: Blocks.ai – control plane and network layer for agents
Blocks.ai provides control plane and network layer infrastructure for connecting AI agents with zero-trust security.
21d ago
Bytechef open source platform for AI agent orchestration and workflow automation
ByteChef is an open-source platform for AI agent orchestration with drag-and-drop workflow building and agentic loop execution.
21d ago
June: Open-source, local-first private AI assistant for macOS with memory
June is an open-source, local-first AI assistant for macOS with voice, memory, and meeting notes in private workspace.
21d ago
Why the Next Era of AI Is About Infrastructure, Not Just Models
Analysis of AI enterprise adoption shift from experimentation to production focus, emphasizing infrastructure, costs, reliability, and data governance.
21d ago
How to feel safe delegating to AI agents?
Discussion on governance frameworks and tools for monitoring and controlling AI agent behavior in production deployments.
21d ago
DeepSeek aims to make its own AI chip
News brief on DeepSeek's plans to develop proprietary AI chips for independent semiconductor manufacturing.
21d ago
Muse Image – new image generation model from Meta
Meta releases Muse Image, an image generation model with faithful instruction following, precise editing, and agentic tool use capabilities.
21d ago
Stacking Chips Sideways Gives AI More Memory
Research on vertically stacked high-bandwidth memory for AI chips increases data density and bandwidth while reducing heat in GPU systems.
21d ago
Altman: GPT-5.6 is 54% more token efficient on agentic coding
OpenAI reports GPT-5.6 Sol model achieves 54% better token efficiency on agentic coding tasks compared to competitors.
21d ago
AI Agent Platforms Are Getting Hacked. Here's What's Missing
Analysis of recent security breaches in Langflow and Dify AI agent platforms highlighting governance and infrastructure gaps.