See how Cyrus compares

Cyrus is an open-source Linear Claude Agent that automates software development workflows. It processes Linear issues assigned to a specific user, creates isolated Git worktrees for each issue, and uses Claude Code to handle development tasks automatically. It's designed to help development teams save time and increase productivity.

Cyrus connects to Linear through the Linear API and webhook system. It listens for issues assigned to a designated user, processes them automatically, and posts Claude's responses back as comments. The agent maintains conversation context and can handle follow-up comments through Linear's webhook API.

Cyrus works with any codebase that Claude Code can understand. This includes most popular programming languages like JavaScript, TypeScript, Python, Java, Go, Rust, and many more. Since it uses Claude Code under the hood, it can work with virtually any text-based programming language or framework.

Basic technical knowledge is helpful for initial setup, but once configured, Cyrus can be used by anyone on your team. You'll need to authorize Linear access, configure repository paths and permissions, and optionally set up MCP (Model Context Protocol) integrations. After setup, team members can simply assign issues to Cyrus through Linear's interface.

Cyrus is designed with security in mind and is SOC 2 Type I compliant. It verifies webhook signatures from Linear, uses OAuth for authentication, and provides configurable tool permissions. Since it's open-source, you can review the code and run it in your own secure environment. All AI interactions are done through Anthropic's secure API, which maintains SOC 2 Type II, HIPAA, and ISO 27001 certifications. For more details, see our agent security documentation.

Yes! Cyrus provides configurable tool permissions. You can specify which tools Claude Code can use when processing issues, such as restricting bash commands with patterns like "Bash(npm:*)" for npm-only access or "Bash(git:*)" for git-only access. This gives you full control over what actions the AI agent can perform in your codebase and helps maintain security.

Cyrus works with any Claude Code plan. For teams handling many Linear issues, higher-tier plans provide better value due to increased usage limits. Check Claude Code pricing for the most current plan options and choose based on your team's expected volume of automated issue processing.

Currently, Cyrus is specifically designed for Linear. However, since it's open-source with a modular EdgeWorker architecture, developers can fork the project and adapt it to work with other issue tracking systems like Jira, GitHub Issues, or GitLab Issues. The core components for session management, Claude Code integration, and workspace creation can be reused.

All of Claude's work is tracked in Git worktrees and Linear comments, so you can review any changes before merging. The agent doesn't automatically merge code - it creates separate worktrees for each issue, allowing human review. You maintain full control over what gets merged into your main branch.

Yes. Cyrus is SOC 2 Type I compliant, ensuring that our security controls and processes meet industry standards for protecting customer data. We implement robust security practices including webhook signature verification, OAuth 2.0 authentication, encrypted data transmission, and granular permission controls. Additionally, Cyrus leverages Anthropic's enterprise-grade infrastructure, which is SOC 2 Type II, HIPAA, and ISO 27001 certified. Our open-source nature allows your security team to audit the codebase, and you can deploy Cyrus in your own secure environment for complete control.

Yes! Cyrus supports Model Context Protocol (MCP) integrations, allowing Claude to connect to your business services and tools. You can configure MCP servers to give Cyrus access to services like Stripe for payment data, BetterStack for monitoring, social media platforms, databases, and many other tools. This enables Cyrus to gather context from your tech stack and make more informed decisions when processing issues.