Sourcegraph alternatives in 2026: 7 code search tools, open source and self-hosted

TL;DR The best Sourcegraph alternative depends on scale and on who searches. For many repositories behind a web UI and an MCP server, pick Sourcebot. For the trigram engine underneath, pick Zoekt. For cross-references in a big tree, pick OpenGrok. For regex over one large repo, livegrep or Hound. For an AI agent that searches code plus docs and data files on one machine, pick XERJ.

Sourcegraph moved to a private monorepo, and its public snapshot names the last commit made under an Apache licence. That sent many teams looking for open-source code search they can host. Background: what semantic code search is.

Summary table

When to choose Sourcegraph instead

Choose Sourcegraph when you want the full commercial product: code search across all repositories, branches and code hosts, code intelligence with references and code owners, and large-scale changes across many repositories at once. Those features are what the alternatives below rebuild in parts.

When to choose Sourcebot instead

Choose Sourcebot for the closest open-source match to Sourcegraph. One setup wizard indexes repositories from GitHub, GitLab, Bitbucket, Azure DevOps or local Git, permission syncing follows your code host, and the web UI searches all of them with regex and boolean queries. Its MCP server gives Claude Code, Codex or Cursor code search, file reading, and symbol definitions and references across repositories you have not checked out.

When to choose Zoekt instead

Choose Zoekt when you want the engine and will build the rest. It is the trigram code search engine Sourcegraph has maintained since 2017, ranks matches with code signals such as symbols, and ships a web server with a simple UI, a JSON API and a gRPC API. Install Universal ctags too, since symbol information is a key ranking signal.

OpenGrok, livegrep and Hound

• Choose OpenGrok for cross-references and version history in a large Java, C or mixed tree. It runs as a Java web app or a Docker container.
• Choose livegrep for instant regex search over gigabyte-scale repositories, the way Google Code Search worked.
• Choose Hound for the lightest setup: a Go backend with a trigram index per repository and a small React UI.
• Choose GitHub code search when the code already lives on GitHub. It searches the default branch and finds symbol definitions.

XERJ: code search for AI agents on your machine

XERJ is a search engine for AI agents. xerj autoindex reads a repository plus the docs, logs, JSONL, CSV, PDFs and SQLite files around it, extracts definitions for 36 tree-sitter languages, and xerj mcp serves 24 tools to Claude Code and other agents. It has no call graph and no web UI for browsing many repositories, and it is single-node. Pick it when one agent needs one index over code and everything beside it. More picks for that job: best MCP servers for code search.

How this page was checked

Licences, install lines and features were read from each project's README, licence file or docs on 2026-11-15, and each one is cited in this page's sources. Star counts came from the GitHub REST API the same day. No head-to-head benchmark was run, so this page makes no speed claims.

Try it with XERJ

Use the Copy the prompt button and paste it into your coding agent. It picks a tool from this table and, if it picks XERJ, indexes your repository and answers one question about the code through MCP.

Read next

Best MCP servers for code search, grep alternatives for AI coding agents, index a monorepo for your agent and syntax-aware search.

FAQ

Is Sourcegraph still open source?

Sourcegraph moved its code to a private monorepo. A public snapshot repository keeps the code as it was before the move and names the last commit made under an Apache licence. The Zoekt search engine that Sourcegraph maintains is still Apache-2.0.

What is the closest open-source alternative to Sourcegraph?

Sourcebot. It indexes repositories from GitHub, GitLab, Bitbucket, Azure DevOps or local Git, gives you a web UI with regex and boolean search, and serves an MCP server with symbol definitions and references. Its licence file is MIT.

Which alternative works with Claude Code or another AI agent?

Sourcebot and XERJ both ship MCP servers. Sourcebot fits many repositories behind a shared server. XERJ fits one repository plus its docs, logs and data files on your own machine. Zoekt has JSON and gRPC APIs an agent can call.

Which one finds references, not only definitions?

Sourcebot's MCP server exposes definitions and references, and OpenGrok is built for cross-references. GitHub code search finds definitions with the symbol qualifier. XERJ extracts definitions in 36 languages and has no call graph.

Do I need Docker?

Sourcebot's setup uses Docker and Docker Compose, and OpenGrok ships a container. Zoekt installs with go install, Hound builds with make, and XERJ installs as one binary.

How was this page checked?

Licences, install lines and features were read from each project's README, licence file or docs on 2026-11-15. Star counts came from the GitHub REST API the same day. No head-to-head benchmark was run.

Evidence

• Sourcegraph transitioned to a private monorepo. sourcegraph-public-snapshot is a public copy of the repository as it was just before the migration, it names the last commit made under an Apache License, and the repository now holds primarily non-OSS-licensed files. Sourcegraph offers code search across repositories, branches and code hosts, code intelligence, and large-scale changes. — https://github.com/sourcegraph/sourcegraph-public-snapshot
• Sourcebot is a self-hosted code context layer for humans and AI agents. It installs with npx setup-sourcebot (Node.js, Docker and Docker Compose), connects GitHub, GitLab, Bitbucket, Azure DevOps or local Git directories, and its MCP server exposes code search, file reading, and symbol definitions and references. Its licence file is MIT, and it lists paid plans. — https://github.com/sourcebot-dev/sourcebot
• Zoekt is a text search engine for source code with substring and regexp matching, based on trigram indexing and syntactic parsing; it ranks with code signals such as symbol matches, recommends Universal ctags, and ships zoekt-webserver with a web UI on port 6070, a JSON API and a gRPC API. Sourcegraph has maintained it since 2017. — https://github.com/sourcegraph/zoekt
• OpenGrok is a source code search and cross reference engine written in Java, understands version control histories of many systems, runs as a Docker container, and is released under CDDL 1.0. — https://github.com/oracle/opengrok
• livegrep provides interactive regex search of gigabyte-scale source repositories, builds with bazel, and runs a codesearch backend plus a web interface. — https://github.com/livegrep/livegrep
• Hound is a code search engine based on a trigram index, with a React frontend and a Go backend that keeps an up-to-date index for each repository; it builds with make and is MIT-licensed. — https://github.com/hound-search/hound
• GitHub code search supports regular expressions, boolean operations, qualifiers and symbol search; it searches only the default branch, excludes files over 350 KiB, and its symbol qualifier finds definitions, not references. — https://docs.github.com/en/search-github/github-code-search/about-github-code-search
• XERJ's tree-sitter registry covers 36 languages and extracts definitions; there is no call graph. — landing/science/syntax-aware-search.html
• On main, xerj mcp registers 24 MCP tools, or 21 when the binary has no Corpus Hub key. — landing/docs/agents/schemas/mcp-tools.json
• GitHub stars on 2026-11-15 from the GitHub REST API: Hound 5,897; OpenGrok 5,027; Sourcebot 4,030; livegrep 2,238; Zoekt 2,069; XERJ 19,111. — https://api.github.com/repos/sourcebot-dev/sourcebot

Related

• Best MCP servers for code search in 2026: 7 compared for Claude Code
• Grep and ripgrep alternatives for AI coding agents: 7 tools compared
• What is semantic code search?
• Does Claude Code index your codebase? How to index a monorepo for your agent
• I need my agent to find a passage and cite the file, not dump the whole tree. How?
• What is syntax-aware search?
• ast-grep vs a code index for an agent?
• How do I set up reference coding so the agent retrieves a mechanism before writing code? · same intent: code search for agents
• Is ripgrep enough for a coding agent, or do I need a code index? · same intent: code search for agents
• How do I audit a PHP codebase with an index instead of grepping 1,400 files? · same intent: code search for agents