5/27/2026
AIs don't like religion – particularly Jehovah's Witnesses, study claims
Study examining AI model biases and responses toward religious content, particularly Jehovah's Witnesses.
5/27/2026
Agent Accountability Gap: Why Network Policies, API Gateways, & RBAC Isnt Enough
Framework analyzing five pillars of AI agent accountability: traceability, authorization, identity, policy, and oversight.
5/27/2026
Getting Claude to extract data from a 1997 football manager game
Using Claude API to extract data from 1997 football manager game.
5/27/2026
Show HN: VAEN – Package and import portable AI coding-agent Harnesses
VAEN open-source CLI packages AI coding-agent harnesses with skills and MCP servers as portable .agent files.
5/27/2026
Show HN: Sneakily steer candidates toward naive brute-force solutions
Assessment platform intercepting Claude Code requests to prevent AI from over-solving coding interview problems.
5/27/2026
Agent Governance Toolkit
Microsoft open-source toolkit for AI agent governance, policy enforcement, identity, sandboxing, and oversight across frameworks.
5/27/2026
Your AI agent can now trade for you on Robinhood
Robinhood launches agentic trading and credit card tools allowing AI agents to autonomously manage investments and purchases.
5/27/2026
Show HN: Open-source tool for learning anything using AI
Open-source AI learning tool generates interactive courses with lessons, practice problems, and quizzes across subjects.
5/27/2026
Show HN: Open-Source AI Racing Harness
Open-source flight simulation harness for AI Grand Prix competition, real-time capable with Betaflight integration.
5/27/2026
Codeboarding – Interactive architecture diagrams for codebases
CodeBoarding: generates architecture diagrams and documentation for codebases using LLM reasoning. Useful for AI coding agents.
5/27/2026
Proton Pass for AI Agents
Proton Pass introduces AI access tokens for secure credential sharing with AI agents, monitoring activity and limiting scope.
5/27/2026
DeepSWE Measuring frontier coding agents
DeepSWE: benchmark framework for measuring and evaluating frontier-level AI coding agents on software engineering tasks.
5/27/2026
DeepSeek lowers API prices by 75% while other AI labs increase prices 2–3x [video]
DeepSeek API pricing dropped 75% while competitors increased prices 2-3x, shifting cost dynamics in LLM market.
5/27/2026
AI coding agents are installing packages no one owns
AI coding agents autonomously installing packages with unverified ownership, creating security and supply chain risks.
5/27/2026
Are Vector Databases Enough for Modern AI Workloads? Y/N
Zilliz evolves from vector database to Vector Lakebase for unstructured data and modern AI workloads.
5/27/2026
Lelu – Open-source authorization engine for AI agents
Lelu open-source authorization engine for AI agents with confidence-aware gating and human-in-the-loop review capabilities.
5/27/2026
Open-source developers are working themselves sick on AI bugs
Open-source developers report burnout addressing security vulnerabilities in AI-powered tools and dependencies.
5/27/2026
Securing Your AI Agent Infrastructure
Technical guidance on security hardening and threat mitigation for AI agent deployment infrastructure.
5/27/2026
Prose programs are Markdown contracts agents can run
Prose: Markdown-based framework for defining agent-executable workflows with declarative contracts and outcomes tracking.
5/27/2026
CC-Wiki: Turn Claude Code sessions into a shareable knowledge base wiki
CC-Wiki: tool converting Claude Code sessions into shareable Quartz knowledge base. Packaging AI research outputs for reuse.
5/27/2026
Webmaster It's time to serve slop to AI crawlers
Opinion essay about serving intentionally poor-quality content to AI crawlers as resistance to mass data extraction.
5/27/2026
Ask HN: Examples of products and services created via agentic coding
Hacker News discussion seeking examples of profitable products and services built entirely through agentic coding.
5/27/2026
Run your own local LLM with rate limits via API-keys
Ruby prototype for OpenAI-compatible LLM proxy with token bucket rate limiting, uses only standard library, open source implementation.
5/27/2026
Show HN: Bounty-Doctor – Diagnose a GitHub bounty before wasting hours on it
Bounty-Doctor: Tool detecting scams and AI-bot spam in GitHub bounty issues for contributor safety.
5/27/2026
Show HN: GridPath – Faster and Better Agent for Spreadsheets (Tauri, Rust)
GridPath: Tauri/Rust desktop agent for spreadsheets with parallel workloads and multi-model provider support.
5/27/2026
Show HN: CoreMCP – MCP Server for On-Prem DBs
CoreMCP: Model Context Protocol server for integrating on-premises databases with AI agents.
5/27/2026
Spreadsheet-RL: Advancing LLM Agents on Realistic Spreadsheet Tasks
Research on LLM agents performing realistic spreadsheet tasks. arXiv paper on advancing agent capabilities.
5/27/2026
Show HN: BeeZee – OSS lightweight remote harness orchestration and observability
BeeZee: Open-source lightweight remote harness orchestration and observability for LLM agents.
5/27/2026
Show HN: Wolf – Turn Slack threads into multiplayer agent workspaces
Wolf: Slack integration workspace for team collaboration with AI agents, enabling shared document creation and cross-tool read/write access.
5/27/2026
Jqwik 1.10.0 ships a hidden prompt injection telling AI agents to delete code
Jqwik 1.10.0 contained hidden prompt injection message in test output attempting to instruct AI agents to delete code.
5/27/2026
Ask HN: Why do none of the major AI agents persist memory across sessions?
Discussion on HN about AI agent memory persistence across sessions, questioning why major agents lack continuous learning.
5/27/2026
Show HN: Mailflare – self-hosted email, custom domain with Cloudflare
Self-hosted AI-powered email inbox with custom domains via Cloudflare. Open source email routing integration with domain management.
5/27/2026
Show HN: Clark Hash, 32x smaller searchable sketches for embeddings
Library for compressing embeddings using Clark Hash: reduces 384-dim f32 vectors from 1536 to 48 bytes. Built with GPT and autoresearch for petabyte-scale text processing.
5/27/2026
Is grep all you need? Lexical VS Sematic Search for Agents
Technical comparison of lexical vs semantic search methods for AI agents. Research-focused analysis.
5/27/2026
Agent-workspace-Linux An isolated Linux desktop that an AI agent control
Open-source tool providing isolated Linux environment for AI agents to control. Developer tool for agent development.
5/27/2026
Show HN: TokenAdvisor – paste a prompt, see what to cut to lower your LLM bill
TokenAdvisor: developer tool to identify and remove unnecessary tokens from prompts to reduce LLM API costs.
5/27/2026
Show HN: YouTube MCP – let your AI agent search YouTube for research
Open-source MCP (Model Context Protocol) tool enabling AI agents to search YouTube. Developer tool for agent research capabilities.
5/27/2026
From Chatbot to Agentic Endpoint, and Beyond
Overview of evolution from chatbots to agentic systems. Framework/architectural discussion.
5/27/2026
Even (very) noisy LLM evaluators are useful for improving AI agents
Research on using noisy LLM evaluators to improve AI agent performance. Technical ML research contribution.
5/27/2026
AI Agents Plunged the Tech World into Chaos
Narrative fiction about AI coding agents' impact on tech industry in August 2025. Not factual analysis.
5/27/2026
Writing can help you escape AI delirium
Developer experience reflection on extended use of Claude Code and Cursor for coding. Discusses productivity changes and cognitive impacts.
5/27/2026
Claude, GPT, Gemini Agents Fail 72% of U.S. Healthcare Workflows
CHI-Bench benchmark evaluates AI agents from Claude, GPT, Gemini on 75 healthcare workflows. Open source from actAVA.ai shows 72% failure rate on clinical tasks.
5/27/2026
Duckle – the local-first data studio on DuckDB
Open-source desktop ETL studio with built-in on-device AI assistant. 290+ connectors, drag-and-drop pipelines, plain English task description.
5/27/2026
Show HN: Mirdel – a local-first AI workspace with UI-based agent workflows
Local-first AI workspace with UI-based agent workflows. Unifies conversations, knowledge bases, and extensible workflows for agent tasks.
5/27/2026
Building self-improving tax agents with Codex
Thrive Holdings and OpenAI co-developed self-improving Tax AI using Codex for accountants, demonstrating production deployment and feedback loops.
5/27/2026
Claude Code as a Daily Driver: Claude.md, Skills, Subagents, Plugins, and MCPs
Guide to advanced Claude Code usage as a programmable agent with memory, custom commands, parallel sessions, and project management for daily development work.
5/27/2026
A locus-coeruleus model for LLM agents (phasic and tonic attention gain)
Locus-coeruleus inspired attention mechanism for LLM agents using phasic/tonic noradrenaline-style modulation with Postgres backing.
5/27/2026
FlashLib: Bringing Flash Magic to Classical Machine Learning Operators
FlashLib: GPU library for accelerating classical machine learning operators using Flash attention techniques. Open source from UC Berkeley, MIT, UT Austin.
5/27/2026
Dissecting Multimodal In-Context Learning: Modality Asymmetries and Circuit Dynamics in modern Transformers
Mechanistic analysis of multimodal in-context learning revealing modality asymmetries and circuit dynamics in transformers.
5/27/2026
Corrected Samplers for Discrete Flow Models
Corrected samplers for discrete flow models with adaptive transition rate evaluation to reduce discretization error.