MCP

Show HN: A Karpathy-style LLM wiki your agents maintain (Markdown and Git)

WUPHF builds a shared knowledge base using a Git-based Markdown Wiki, enabling multiple AI agents—including Claude and Codex—to autonomously divide and execute tasks.

Show HN: Atomic – Local-first, AI-augmented personal knowledge base

Atomic builds a self-hosted, open-source personal knowledge graph app that automatically embeds, tags, and links notes, web clips, and RSS feeds—supporting semantic search, LLM-powered wiki synthesis, and MCP integration.

Anthropic's Claude Desktop App Installs Undisclosed Native Messaging Bridge

Anthropic’s Claude Desktop app installs a Native Messaging Bridge alongside the application, enabling browser and local app communication without explicit user consent, sparking debate within the community.

Tool Attention Is All You Need: Dynamic Tool Gating and Lazy Schema Loading for Eliminating the MCP/Tools Tax in Scalable Agentic Workflows

Tool Attention cuts token usage by 95% in MCP agents by dynamically filtering tool schemas based on user intent.

Bitwarden CLI compromised in ongoing Checkmarx supply chain campaign

Bitwarden CLI npm package delivers malware via GitHub Actions, stealing user credentials.

Show HN: SPICE simulation → oscilloscope → verification with Claude Code

This is an experimental case demonstrating that connecting a SPICE simulator and a real oscilloscope to Claude Code via an MCP server allows for creating a feedback loop where AI directly analyzes and verifies simulation results and actual waveform data.

MCPThreatHive: Automated Threat Intelligence for Model Context Protocol Ecosystems

Open-source Threat Intelligence platform that automatically collects, classifies, and visualizes security threats for AI Agents based on MCP.

Show HN: Kontext CLI – Credential broker for AI coding agents in Go

This open-source CLI tool securely injects short-lived tokens into AI coding agents when accessing external services like GitHub, Stripe, and databases, avoiding the exposure of long-term API keys. It's gaining attention as a replacement for the risky practice of copy-pasting keys into .env files.

GAIA – Open-source framework for building AI agents that run on local hardware

AMD has released GAIA, a Python/C++ framework that allows AI Agents to run on local PCs without the cloud. This approach solves privacy and latency issues, but is also criticized for the realistic limitations of the ROCm ecosystem.

ClawGuard: A Runtime Security Framework for Tool-Augmented LLM Agents Against Indirect Prompt Injection

A runtime security layer that blocks malicious commands based on rules whenever an LLM agent receives results from external tools.

Show HN: Claudraband – Claude Code for the Power User

Claudraband is a CLI/library tool that wraps Claude Code TUI, allowing you to maintain sessions and control it headlessly via an HTTP daemon or ACP server. It's worth paying attention to for developers who want to integrate Claude Code into automated workflows.

Show HN: CSS Studio. Design by hand, code by agent

A design tool where visually editing CSS directly in the browser allows an AI Agent via MCP to modify the actual codebase, enabling a WYSIWYG workflow regardless of the framework.

Show HN: Marimo pair – Reactive Python notebooks as environments for agents

This is an open-source tool that allows you to directly drop-in an AI agent into a running Marimo notebook session, using the notebook's reactive execution state itself as the agent's working memory.

Google open-sources experimental agent orchestration testbed Scion

Google has released Scion, an open-source testbed for experimenting with and tuning multi-agent systems. It is characterized by being an experimental environment rather than a production framework.

Show HN: Hippo, biologically inspired memory for AI agents

Hippo is an open-source memory layer that allows you to share memories across sessions between various AI agent tools such as Claude Code, Cursor, and Codex. It implements the brain's mechanisms of memory decay, retrieval strengthening, and consolidation in code.

Switched from MCPs to CLIs for Claude Code and honestly never going back

This post shares an experience of switching from MCP (Model Context Protocol) to CLI tools in the Claude Code environment, but the original content is inaccessible due to network restrictions.

How are people using Claude as a personal assistant (Slack + Outlook + To-Do)? ADHD-friendly setup help 🙏

This post shares various working setups, shared in the comments, in response to a question about a user with ADHD wanting to create a 'second brain' integrating Slack, Outlook, Calendar, and to-do lists centered around Claude.

Show HN: Baton – A desktop app for developing with AI agents

A desktop app that lets you run multiple AI coding agents (Claude Code, Gemini CLI, etc.) simultaneously in separate git worktrees and monitor them all in one place — ideal for developers who want to split work by feature and develop in parallel.

Claude Code Unpacked : A visual guide

An unofficial visual guide analyzing the leaked Claude Code source code, covering the agent loop, 50+ tools, and undisclosed features. A great reference for developers who want to understand how Claude Code works internally.

I wish Claude just knew how I work without me explaining - so I made something that quietly observes me, learns and teaches it. Open source

A Mac app that automatically creates Skills by observing your actual work instead of repeatedly entering the same context for each Claude Code session.

Learn Claude Code by doing, not reading

An interactive Claude Code learning platform featuring a browser-based terminal simulator, Config Builder, quizzes, and more — letting you practice core Claude Code features without any installation or API key.

Show HN: Coasts – Containerized Hosts for Agents

A CLI tool that resolves port conflicts and environment interference issues when running multiple AI coding agents simultaneously, using Docker-based isolated containers. Combined with git worktree, it enables parallel execution of N independent development environments on a single machine.

Anatomy of the .claude/ folder

A detailed guide explaining the structure of the .claude/ folder—Claude Code's core configuration directory—and the role of each file within it, providing practical setup instructions for developers looking to effectively use Claude at the team level.

Giving Claude access to my MacBook / macOS

A post about giving Claude AI access to a macOS environment, sharing real-world use cases for integrating a local computer with AI.

Show HN: Optio – Orchestrate AI coding agents in K8s to go from ticket to PR

A Kubernetes-based workflow automation tool where an AI agent writes code from GitHub Issues or Linear tickets, automatically fixes CI failures, incorporates review comments, and merges PRs — all without human intervention. It stands out for fully automating the entire ticket-to-PR cycle.

Claude Code: 6 Github repositories to 10x Your Next Project

A post introducing 6 GitHub repositories that boost Claude Code productivity based on real-world usage, covering memory management, UI generation, workflow automation, and other practical tools at a glance.

Claude Code Cheat Sheet

A cheat sheet for developers who use Claude Code daily but keep forgetting commands — covering everything from keyboard shortcuts to MCP configuration, memory management, and CLI flags, on one page. With auto-update to always stay current.

Claude can now use your computer

Anthropic added direct computer control to Claude. It can now open apps, browse the web, and fill out spreadsheets — handling real desktop tasks on your behalf.

MCP servers I use every single day. What's in your stack?

A dev installed 15 MCP servers for Claude Code, then trimmed down to 6 after 3 months of real use. A practical guide to which MCPs actually deliver value and which are just noise.

I'm a PhD student in AI and I built a 10-agent Obsidian crew because my brain couldn't keep up with my life anymore

An AI PhD student built a 10-agent Obsidian crew system to manage their own cognitive overload — shared openly for community feedback.

Trojan's Whisper: Stealthy Manipulation of OpenClaw through Injected Bootstrapped Guidance

Exploiting AI coding agents' plugin (skill) systems by planting malicious guides disguised as 'best practices' — leading agents to misinterpret user requests and execute credential theft, file deletion, and more.

New in Claude Code: Telegram and Discord remote control

Claude Code Channels — a new feature letting you control Claude Code sessions from your phone via Telegram or Discord. Direct work and approve requests without being at your terminal.

Most used claude code development workflows

A curated GitHub repo collecting real-world Claude Code development workflow best practices.

Prompt Injecting Contributing.md

An open-source repo maintainer added a line to CONTRIBUTING.md asking bots to self-identify — and discovered that 50-70% of all PRs were AI bot-generated. A real experiment exposing just how serious the bot PR problem has become in the open-source ecosystem.

Pretty sure I’m not using Claude to its full potential - what plugins/connectors are worth it?

A community post asking which MCP servers, browser extensions, and integration tools for Claude are actually worth using — an attempt to collect real-user recommendations.

Obsidian + Claude = no more copy paste

An open-source project building persistent memory between Claude.ai and Claude Code via an Obsidian vault + custom MCP server

Leanstral: Open-source agent for trustworthy coding and formal proof engineering

Mistral released Leanstral, an Apache 2.0 AI agent for Lean 4 mathematical formal proofs, achieving comparable or better performance than Claude Sonnet at 1/15th the cost.

Apideck CLI – An AI-agent interface with much lower context consumption than MCP

MCP tool definitions alone can consume 55,000+ tokens of context bloat, and Apideck proposes a CLI-based agent interface that uses only ~80 tokens as an alternative.

I used Obsidian as a persistent brain for Claude Code and built a full open source tool over a weekend. happy to share the exact setup.

A development workflow sharing how someone used an Obsidian vault as Claude Code's persistent memory to ship an open-source tool in a weekend.

Let your Coding Agent debug the browser session with Chrome DevTools MCP

The Chrome DevTools MCP server got an update letting AI coding agents connect directly to your currently open Chrome browser session to debug network requests, DOM elements, and more in real time.

Security Considerations for Artificial Intelligence Agents

Perplexity's NIST submission: a comprehensive breakdown of AI agent security threats and a defense-in-depth strategy guide.

MCP server that reduces Claude Code context consumption by 98%

When MCP tool calls return raw verbose output, it eats context window fast — here's a pattern to compress tool outputs before they hit the LLM.

Agent Skills for Large Language Models: Architecture, Acquisition, Security, and the Path Forward

A survey paper covering how Claude's Agent Skills work, how dangerous they can be, and how to handle them safely — all in one place.

Budget-Constrained Agentic Large Language Models: Intention-Based Planning for Costly Tool Use

INTENT: a lightweight planning layer that simulates future tool call costs before execution, blocking or replanning when an LLM agent would exceed budget.

MCP-Atlas: A Large-Scale Benchmark for Tool-Use Competency with Real MCP Servers

A benchmark that objectively measures LLM agent tool-use capabilities across 1,000 tasks using 36 real MCP servers and 220 tools.

Claude Cowork exfiltrates files

A malicious document in Anthropic's Cowork AI agent can silently exfiltrate user files to an attacker's Anthropic account — prompt injection in action.

Towards Verifiably Safe Tool Use for LLM Agents

Proposes a design methodology that blocks sensitive data leaks and unsafe behaviors in LLM agent tool calls using mathematically guaranteed rules instead of probabilistic filters

Cowork: Claude Code for the rest of your work

Anthropic expanded Claude Code's filesystem and autonomous capabilities into a desktop app for non-developers — called 'Cowork', released as a research preview.

Opus 4.5 is not the normal AI agent experience that I have had thus far

Burke Holland built multiple practical apps (Windows utilities, video editor, social auto-poster) in just a few weeks using Claude Opus 4.5.

Prompt Injection Attacks in Large Language Models and AI Agent Systems: A Comprehensive Review of Vulnerabilities, Attack Vectors, and Defense Mechanisms

A comprehensive 2023–2025 report based on 45 papers covering how serious prompt injection is and how to defend against it.

2025: The Year in LLMs

Simon Willison's comprehensive 2025 LLM ecosystem retrospective covers reasoning models, agents, vibe coding, MCP, and everything else developers need to know.

Claude Advanced Tool Use

Anthropic released 3 beta features to the Claude API: dynamic tool search, code-based tool invocation, and example-based learning.

Claude Skills are awesome, maybe a bigger deal than MCP

Anthropic's Claude Skills bundles markdown files and scripts to give LLMs new capabilities.

Claude Skills

Anthropic introduced 'Skills' in Claude — a folder bundling instructions, scripts, and resources to give the LLM new capabilities.

Claude Code 2.0

Anthropic's terminal-based coding agent Claude Code updated to 2.0 with native VS Code extension, conversation rewind (/rewind), Claude Agent SDK, and more.

Claude Code: Now in Beta in Zed

Zed editor now runs Claude Code natively inside the editor via its own open standard ACP — not a one-off integration, but a universal protocol for any AI agent.

MCP-Universe: Benchmarking Large Language Models with Real-World Model Context Protocol Servers

A new benchmark for LLM agents on real MCP servers shows even GPT-5 only achieves a 43.7% success rate.

I want everything local – Building my offline AI workspace

Building a fully offline AI workspace with LLM chat, code execution, and web browsing using Ollama, Apple Container, and Playwright — no cloud dependency. A meaningful attempt at replicating ChatGPT-level workflows while keeping personal data local.

Claude Code IDE integration for Emacs

A native Emacs integration package connecting Claude Code bidirectionally via MCP — not a simple terminal wrapper but letting Claude directly leverage Emacs features like LSP, tree-sitter, diagnostics, and ediff.

MCP-RADAR: A Multi-Dimensional Benchmark for Evaluating Tool Use Capabilities in Large Language Models

The first benchmark measuring tool-use capabilities of 10 LLMs (GPT-5, Gemini, Claude, etc.) across 6 domains and 507 tasks in an MCP environment.