Octocode, developed by Muvon, is an MCP server tool built with Rust, designed for code indexing and semantic search. It connects to AI assistants and code repositories, both local and GitHub-hosted, building intelligent knowledge graphs of codebases. Distributed as an npm package and through one-click installers for compatible AI clients, Octocode functions as a Model Context Protocol server. It enhances developer and agent workflows by providing deep insights into code structure and relationships. This free tool makes advanced code analysis capabilities accessible for diverse engineering tasks.

What does Octocode connect to?

Octocode connects to AI assistants and agents, alongside both local and GitHub repositories, encompassing public and private codebases. It operates as an MCP server, functioning as a specialized code indexer and semantic search engine. This setup allows Octocode to construct intelligent knowledge graphs of codebases, providing a structured understanding of software projects. Its primary role is to bridge AI capabilities with code intelligence, enabling deeper analysis and interaction with various development environments and code sources, enhancing developer productivity and understanding.

Core features and tools

Octocode exposes several key tools via its MCP server. These include semantic_search for semantic code search, and view_signatures to inspect file signatures and code structure. The graphrag tool enables live file/symbol graph operations like search, get-node, and find-path. For AST-based analysis, structural_search is available. Additionally, lsp_* tools provide LSP integration for typed semantic navigation, covering definition, references, and callHierarchy. Built-in commands such as research, plan, review_pull_request, and generate support common development workflows effectively.

Who is it for?

Octocode targets developers, product managers, analysts, engineering managers, indie developers, and AI agents. It is designed for anyone needing deep code insights, whether for strategic planning, detailed code investigation, or automated development tasks. The tool's capabilities support roles requiring a granular understanding of codebases and their interconnections. This is not the right fit if your workflow does not involve connecting to AI clients or analyzing code repositories.

How to get started

Access is granted through a standard OAuth authorization flow for GitHub-backed tools, initiated via "npx octocode auth login". The MCP server supports "stdio" as a transport protocol, and "HTTP" when using the "octocode mcp --multi" command.

Use cases

• Deep code forensics and research, such as understanding internal frameworks by validating assumptions against real code using semantic search and import tracing.
• Evidence-based engineering workflow for planning, involving researching patterns in quality repositories before planning and implementing, and coordinating parallel agents for multi-repository complexity.
• Defects-first PR review, focusing on bugs, security, architecture, and performance across changed code, generating prioritized fixes with context from local and main/changed branches.

Octocode provides a powerful MCP server for connecting AI agents with codebases, offering advanced indexing and search capabilities. Its array of tools supports detailed code analysis and workflow automation for various engineering roles. The platform's ability to build intelligent knowledge graphs aids in understanding complex projects.