Yaw Labs

Claude Fable 5.1, GPT-6 Astra and Gemini 3.8 Flash in four days - Nvidia's $12.93B Hugging Face deal https://lnkd.in/gHhQnu2U

Prod Bugs https://lnkd.in/gAKrCVts

SiliconSnark wrote an article about what we're building at Yaw Labs. Worth a read if you want a clear outside take. A few themes from the piece that match how we think about the work: The terminal is no longer a lonely rectangle. It's shared space between developers, remote sessions, databases, and coding agents. We build for that reality. What we ship: Yaw Terminal. A desktop terminal for Windows, macOS, and Linux. Split panes, an integrated editor, SSH, database connections, and AI assistance that pairs with tools like Claude Code. Bring your own keys. Traffic stays between you and your model provider. mcp.hosting (now Yaw MCP) One managed endpoint for your MCP servers. The ecosystem is moving fast, and the servers are a plenty, so we centralized & streamlined it. All local & free. typed.cloud. Monthly billing and lower overage rates for teams running heavy on Claude Code. Pricing is what we say it is. No hidden subscription games. Our orchestrators, compliance suites, and MCP servers (AWS, Tailscale, SSH, npm, and others) are open source. Trust gets earned, not announced.   https://lnkd.in/eymwHm9q

@yawlabs/postgres-mcp - a maintained Postgres MCP server, since Anthropic's reference one is deprecated (and has an unpatched SQLi) Anthropic's reference Postgres MCP server was archived in May 2025 and deprecated on npm a couple months later. No replacement ever shipped. It still pulls ~450,000 downloads a month. That's a lot of AI agents running on unmaintained code - code with an unpatched stacked-query SQL injection. The BEGIN READ ONLY wrapper it relies on does nothing against: SELECT 1; COMMIT; DROP TABLE x; because the simple query protocol runs every statement in the string. Read-only or not, the DROP executes. So I built a replacement: @yawlabs/postgres-mcp. MIT, open source. It fixes the injection structurally - all user SQL goes through Postgres's extended query protocol, which allows exactly one statement per request, so the database itself rejects multi-statement payloads at the wire level. There's an integration test that asserts it. It also closes a few other footguns in the original: - App-side row limits that could OOM the process - now capped before the result set is materialized. - EXPLAIN ANALYZE that silently mutated data - now rolls back, and writes are opt-in anyway. - No pool or timeout caps - now configurable, with sane defaults. And it's a daily driver, not just a query box: 19 tools covering schema introspection, EXPLAIN plans, slow-query diagnostics, lock inspection, index advice, roles and privileges, replication status. It's a drop-in replacement - swap the package name, move your connection string to a DATABASE_URL env var, done. Works with Claude Code, Cursor, and any MCP client. If you've got agents touching a production database, it's worth knowing what's running underneath them. Repo + full writeup of the bug and the fix in the comments. #PostgreSQL #MCP #AI #ClaudeCode #DatabaseSecurity #DeveloperTools

First demo of Yaw Terminal - a desktop terminal built for people who already live in Claude Code (or want to). What's in the video: - Multi-pane: one CC session per pane, all visible at once - Broadcast mode: type once, send the same input to every pane (great for /fast, /model, /clear, or any prompt you want to fan out across parallel agents) - Restore tab: closed a session by mistake? Bring it back with ease! - The flow of running 3-5 Claude Code agents in parallel without losing your place For Claude Code power users who've outgrown one-session-per-tab.

Code w/ Claude doubled Claude Code's rate limits - AWS MCP Server hit GA the same day - we shipped two more Production books This was a reset week on AI coding infrastructure - the kind that moves defaults, not just feature flags. Tuesday May 6 was Code w/ Claude in San Francisco. Anthropic doubled Claude Code's 5-hour rate limit for Pro, Max, Team, and Enterprise, and removed the peak-hours penalty entirely. Opus 4.7 was confirmed generally available at the same pricing as 4.6. Managed Agents picked up multi-agent orchestration, outcome targets, and "Dreaming" - a cross-session memory loop where the model reviews finished sessions and writes new memories before the next one starts. The capacity story is bigger than the model story: Anthropic announced it is taking the entirety of SpaceX's Colossus 1 in Memphis, 300MW and roughly 220,000 NVIDIA GPUs, which is what funds the rate-limit doubling. The same Tuesday AWS announced the AWS MCP Server going GA - free, IAM-gated, with sandboxed Python execution against any AWS API. MCP just crossed from research moment to default cloud primitive. The hygiene chapter showed up alongside it: CVE-2026-33032 (CVSS 9.8) lets unauthenticated attackers take over nginx-UI MCP endpoints, and a separate design flaw in the MCP STDIO transport allows arbitrary OS command execution with up to ~200,000 servers in scope. https://lnkd.in/gJj7XVyu

Yaw Mode is an opinionated config overlay for Claude Code. You toggle it on and the next session inherits a curated bundle of rules, skills, and agents without touching your ~/.claude. Turn it off and you're back to vanilla Claude Code, same conversation history, nothing mutated. The rules are mostly things I got tired of re-explaining to Claude: no speculative abstractions, validate only at system boundaries, never bypass branch protection, integration tests hit real DBs, etc. Skills and agents are bundled alongside. Link has the full breakdown of what's inside and how the overlay keeps your conversation history safe. https://lnkd.in/geP284yb

mcp.hosting – The easiest way to install MCP servers Adding new mcp servers by hand-editing JSON across Claude Code, Claude Desktop, and Cursor is annoying. So I built https://mcp.hosting the easiest way to install MCP servers. Add mcp servers by clicking to add from the Explore page. Or click on github repo badges. Or manually add as well. It's easy to add a bunch in your online account and then they're immediately available in your mcp client of choice. There is also Smart Routing built in to make sure it's fast and uses the best mcp tool for the job. Free tier covers 3 active servers, Pro is $9/mo for unlimited, and self-host is available if you want to run the whole stack.

Excited to introduce yaw labs. The team behind http://yaw.sh. We're building the terminal for the AI era. Yaw is a full-featured terminal with native support for Claude Code, OpenAI Codex, Gemini CLI, and more. All in one place! No more switching between tools, tabs, or providers. Just your workflow, unified. We're early, we're building fast, and we're looking for developers who want to shape what comes next. If you're a developer, AI enthusiast, or just someone who lives in the terminal. We'd love to connect!

Four AI coding agents now run directly in your terminal — Claude Code, Codex, Gemini CLI, and Vibe CLI. I put together a practical comparison: what each one does, how to install them, and which to start with based on your setup. The short version: if you're still switching to a browser to talk to an AI about your code, you're leaving speed on the table. https://lnkd.in/gaR7yzfY