OpenBug is an AI-powered debugging CLI designed for developers who need real-time insights into running applications. It combines log capture, codebase search, and an interactive AI assistant into a single tool that runs in the terminal. The core value is eliminating the constant context-switching between logs, code editors, and separate debugging tools. Instead, developers can ask plain-language questions about their system and get answers derived from actual runtime data. This tool is especially useful for debugging microservices, understanding unfamiliar codebases, or tracing issues across multiple services without leaving the command line.
The primary pain point OpenBug solves is the inefficiency of traditional debugging workflows where developers manually grep through logs scattered across multiple terminals, then cross-reference code locally. When a production issue like an auth endpoint failure arises, the developer must switch between error logs, code files, and documentation, hoping to connect the dots. OpenBug eliminates this fragmentation by capturing logs automatically and making them accessible alongside the codebase in a chat interface. This matters because debugging time is often wasted on context restoration rather than actual analysis, and OpenBug collapses that process into a single conversation.
The first major feature is automatic log capture via the `debug <command>` prefix. When a developer runs `debug npm run dev` or `debug python app.py`, OpenBug intercepts the process, streams logs to a local cluster, and makes them available to the AI assistant in real time. Unlike manual log shipping, this requires no configuration beyond installing the npm package. The benefit is immediate: developers see logs flow normally in their terminal while simultaneously enabling the AI to reference those logs when answering questions. This makes debugging dynamic rather than retrospective.
The second set of features focuses on natural language codebase search and AI chat. Developers can ask questions like "Why is the auth endpoint failing?" or "Where do we handle payment webhooks?" directly in the assistant terminal. The AI agent searches the local codebase (which stays on disk) and correlates the relevant code sections with the captured logs. Unlike coding assistants like Cursor or Copilot that only do static analysis, OpenBug sees what the code actually produced at runtime. This is particularly powerful for diagnosing race conditions, config errors, or schema drift that don't appear in the source code alone.
admin
A third major capability is multi-service debugging across a local cluster. The `debug` command launches a cluster server on `ws://127.0.0.1:4466` that multiple service terminals can connect to. For example, one terminal runs the backend with `debug npm run dev`, another the frontend with `debug npm start`, and the AI assistant in a third terminal sees logs from both services simultaneously. The AI can then trace requests across services—for instance, identifying that a login failure originates from a JWT validation issue in the authentication microservice, not the frontend. Self-hosting is also supported by configuring a custom WebSocket endpoint.
The overall workflow operates through a two-terminal architecture. Terminal 1 runs `debug` to start the AI assistant, which connects to the local cluster. Terminal 2 runs `debug <command>` to start the service, which also connects to the same cluster and streams logs. The cluster then communicates with OpenBug's AI server via WebSocket, where an Agent Graph processes queries: it analyzes logs, searches the codebase, and generates insights. Responses flow back through the cluster to the assistant terminal. The architecture is designed to support multiple services and scale to complex applications without requiring changes to the service code itself.
Concrete use cases range from debugging microservices with real bugs to understanding a new codebase. The interactive demo at github.com/openbug-ai/demo showcases three realistic bugs (schema drift, config errors, race conditions) across two microservices, demonstrating how OpenBug correlates logs and searches code across services. In a typical workflow, a developer might ask "Users can't log in, what's wrong?" and the AI will show logs from the last auth request, identify the failing JWT validation, and even suggest fixes based on the codebase context. For legacy code, asking "Where do we validate payment status?" returns the exact file and function, all without leaving the terminal.
OpenBug targets developers working with Node.js 20+ tools (npm, yarn, bun) who debug microservices, cloud apps, or any command-line-run software. It is currently in free beta with an MIT license, allowing self-hosting or use via the hosted service at app.openbug.ai. The tool is particularly valuable for backend engineers, DevOps engineers, and full-stack developers who need to debug cross-service issues quickly. By turning debugging into an AI-powered conversation that sees both logs and code, OpenBug transforms a time-consuming, manual process into an interactive, efficient experience.
OpenBug is built for software developers, DevOps engineers, and full-stack developers who work with Node.js 20+ applications and need to debug running services in real time. It is especially valuable for teams debugging microservices architectures where issues span multiple services, as the tool can capture logs from all connected services and trace across them. Backend developers debugging API endpoints, frontend developers troubleshooting integration issues, and engineers maintaining legacy or unfamiliar codebases will all benefit from the natural language codebase search and AI-powered log analysis. The tool also suits technical leads and senior engineers who need to quickly understand and resolve production issues without deep knowledge of every service.