Pushci CLI, developed by Finsavvyai, is a free Model Context Protocol (MCP) server designed for AI-native CI/CD operations. Built primarily with Go, it provides a governed runtime for AI agents to manage development pipelines. Distributed through various methods, including npm, Homebrew, and direct script execution, it ensures a consistent, signed binary across platforms. This tool acts as a crucial delivery boundary, connecting AI agents with CI/CD operations. It supports both human developers and AI clients like Claude Code and OpenAI, enabling advanced pipeline automation and management.

What does Pushci CLI connect to?

Pushci CLI functions as an MCP server, establishing connections with AI agents for CI/CD tasks. It serves as a delivery boundary, offering a controlled runtime environment for these agents to execute operations. The tool supports the "stdio" transport protocol for local MCP server interfaces, while remote interactions occur via HTTP to "api.". It is compatible with several AI clients, including Claude Code, Cursor, Windsurf, OpenAI, and Gemini, enabling a broad range of AI-driven development workflows.

Core features and tools

Pushci CLI provides a set of specific callable tools to aid in CI/CD management. These include pushci_doctor, which diagnoses environmental issues within the setup. pushci_diagnose offers AI-powered analysis for errors encountered in pipelines. For security, pushci_scan performs security scans of pipelines. Additionally, pushci_recommend suggests best-fit deployment targets and integrations. Finally, pushci_promote registers pipelines with AI registries and search engines, enhancing visibility and discoverability for AI agents.

Who is it for?

Pushci CLI is designed for both human developers and AI agents involved in CI/CD processes. Developers can use it for local verification and zero-config setup, while AI agents can autonomously manage and promote pipelines. Its capabilities extend to enterprise environments through SSO support. The tool's focus on AI-native CI/CD makes it suitable for teams adopting advanced automation. Skip this if your CI/CD workflow does not involve AI agents or require local-first verification.

How to get started

The authentication model involves JSON Web Tokens with a one-year expiry, rotated on request, saved locally after a browser-based "pushci login" command. This method ensures secure access for human users, while the local MCP server interface supports the "stdio" transport protocol for direct communication with compatible AI clients, requiring specific configuration within their "mcpServers" entry.

Use cases

• Zero-config CI/CD setup: Scans repositories, detects languages and frameworks, then generates a "pushci.yml" file quickly.
• Local-first verification: Allows developers to build, test, scan, cache, and validate changes locally before deploying to remote compute.
• AI-driven pipeline management: Enables AI agents to autonomously plan, run, diagnose, and promote pipelines, such as deploying services to staging environments.

Pushci CLI offers a specialized solution for integrating AI agents into CI/CD workflows, providing a governed environment and essential tools. Its support for various distribution methods and authentication models makes it accessible for diverse teams. The focus on local verification and AI-driven automation positions it as a valuable asset for modern development practices. Try using "pushci_doctor" to diagnose environment issues in your setup.