You're in a Slack thread. You type @cyrus. It answers. Instantly. In the thread. Because it's already connected to your Linear, your GitHub, your repos. No new tools. No new workflows. It just plugs in.
Your PRs shouldn't sit for days waiting for a review. You can now @mention Cyrus directly on any GitHub pull request and get an instant, thorough code review — edge cases caught, architectural trade-offs questioned, concrete improvement suggestions.
That feature that was going to wait till next week? It ships before lunch. Cyrus now runs Cursor Composer (250 tokens/sec) and OpenAI Codex (77.3% Terminal-Bench accuracy) — giving you the speed and flexibility to choose the right model for each task.
Your AI coding agent can write code. But can it generate a PowerPoint deck, follow your team's conventions, or debug systematically? Skills are markdown files that teach Claude Code new capabilities - and the Skills Marketplace makes discovering them effortless.
Anthropic's most capable coding model is here. Opus 4.6 brings a 1M token context window, agent teams, and adaptive thinking - and it's now the default for every Cyrus agent. Here's what it means for your team.
Anthropic just announced swarm mode alongside Opus 4.6 - turning Claude Code from a single AI coder into a team lead that spawns agent teams of specialists. Here's everything you need to know about Claude Code agent teams and the multi-agent feature that changes how we build software.
MCP tool definitions were consuming 66,000+ tokens before you typed anything. Tool Search fixes this by loading tools on-demand, reducing token overhead by 85%. Here's everything you need to know about the feature that makes large MCP setups finally practical.
Claude Code tends to produce verbose, over-engineered code. The code-simplifier agent is an official Anthropic plugin that cleans up AI-generated code automatically - preserving functionality while eliminating unnecessary complexity. Here's how to use it.
Security has always been foundational to how we build Cyrus. Today, we're proud to announce SOC 2 Type I compliance, independently verified by Prescient Security. Here's what this means for our users.
What if Claude just kept going? The Ralph Wiggum technique creates autonomous development loops where Claude iterates on its own work until the task is genuinely finished. Here's everything you need to know about the plugin developers can't stop talking about.
You're refactoring your authentication system. Cursor forgets context halfway through. Claude Code handles the entire codebase. Here's why developers are switching - and when you shouldn't.
The AskUserQuestion tool is transforming how developers work with Claude Code. Instead of making assumptions, Claude interviews you to build a spec before executing. Here's how to use it - and how Cyrus brings this capability directly into Linear.
A complex refactor that would take a senior engineer 3 weeks completed in 3.5 hours - zero human intervention. Watch how an orchestrator agent coordinated six specialized agents, verified their work, and delivered a complete refactor. This is orchestration.
You're building a dropdown menu. You test it manually. It breaks on mobile. You fix it. Repeat. What if AI could do all of that - navigate, test, screenshot, and fix issues automatically? Here's how Playwright MCP transforms UI development from repetitive manual work into intelligent automation.
A coffee break. That's all it took. While I sipped my espresso, Cyrus resolved a customer's feature request, opened a PR, and updated Linear. Here's how we closed 38 issues in 3 weeks with almost zero intervention.
