What is
Sourcegraph
?
Most platform initiatives fail the same way: they start too big, get funded like projects instead of products, build for too long without testing with users, and confuse renaming an infrastructure team with building a platform team. Instead, start small, learn fast and scale deliberately.
At a glance
Fact | Value |
|---|---|
Primary use case | Codebase-wide search and intelligence that provides context to developers and AI agents |
Deployment type | Web-based platform with hosted and self-hosted deployment, including Kubernetes |
Pricing model | Enterprise quote scaled by team size; outcome-based pricing per merged changeset for Agentic Batch Changes; free public-repository search |
API compatibility | Model Context Protocol (MCP), SCIP code intelligence, streaming search API using SSE, webhooks |
Supported integrations | GitHub, MCP-compatible coding agents |
Source type | Dual |
Cloud provider | Cloud Agnostic |
Ideal team size | Enterprise engineering teams; 50+ developers |
Required skills | Repository and version-control administration, code search and navigation, site configuration; Kubernetes administration for Kubernetes-based self-hosting |
AEP branch | Agent Infrastructure |
AEP category | Harness |
AEP subcategory | Context |
AEP detail | Search & Retrieval |
Profile
Search & Retrieval is Sourcegraph’s role in the AEP taxonomy. Sourcegraph indexes repositories across a codebase to provide code search and intelligence for developers and AI agents. Its capabilities include exact code search, natural-language Deep Search with answers grounded in citations, code intelligence for navigation, and tools for monitoring and changing code across repositories.
The platform includes a Sourcegraph MCP server that provides agents with codebase context, and Agentic Batch Changes for large-scale code changes. Sourcegraph reports that it is trusted by more than 200 enterprise engineering teams; its public search covers more than 2 million open-source repositories.
Focus
Sourcegraph is aimed at enterprise engineering teams and AI agents that need context across large codebases. It addresses the problem of agents seeing only fragments of an enterprise codebase, which can leave cross-cutting changes incomplete. Repository-wide search and code intelligence help teams identify related code and understand system-wide impact before changes ship.
Background
The supplied company-background research does not establish Sourcegraph’s founding date or funding history. It documents Sourcegraph’s open-source contributions: a 2018 report describes the company making the code for its engine and Chrome and Firefox extensions available in a public GitHub repository. The company’s public GitHub presence also includes the Sourcegraph public snapshot repository.
Features
Code Search
Feature: Search code across repositories. Benefit: Find exact, deterministic results across a codebase; Sourcegraph also offers free public search across more than 2 million open-source repositories. Technical execution: Sourcegraph indexes repositories across the codebase and provides code search.
Deep Search
Feature: Ask complex questions about a codebase in natural language. Benefit: Get answers grounded in citations for questions from engineering, support, and go-to-market teams. Technical execution: Deep Search returns answers with citations based on codebase context.
Sourcegraph MCP server
Feature: Provide coding agents with Sourcegraph code intelligence. Benefit: Give agents broader code context to support reliable results with fewer retries. Technical execution: The MCP server provides complete, SCIP-powered context and supports MCP-compatible coding agents.
Code oversight
Feature: Track code changes, patterns, and risk across a codebase. Benefit: Help teams understand system-wide impact before changes ship and stay informed as code evolves. Technical execution: Code Insights track migrations, adoption, and risk over time; Code Monitoring alerts teams and agents when code changes.
Agentic Batch Changes
Feature: Apply code changes across repositories. Benefit: Support migrations, modernization, and remediation at scale with control over the changes. Technical execution: Sourcegraph describes Agentic Batch Changes as an AI agent for large-scale code changes across repositories.
Living Documentation
Feature: Maintain a knowledge base of a codebase. Benefit: Give developers and agents a continuously updated source of codebase context. Technical execution: Sourcegraph describes Living Documentation as a continuously updated knowledge base of the codebase.
Pros and cons
Pros
Indexes repositories across a codebase, helping developers and agents find relevant code beyond the files in a single task.
Combines exact code search with natural-language Deep Search and cited answers.
Provides SCIP-powered MCP context for coding agents and tools for monitoring, code insights, and changes across repositories.
Offers free search across more than 2 million open-source repositories and reports trust from more than 200 enterprise engineering teams.
Cons
Kubernetes-based self-hosting requires Kubernetes administration, in addition to repository and site administration skills.
A 2020 GitHub issue tracked concerns about flaky or confusing alerts on sourcegraph.com. This is a historical issue report and does not establish whether the concern remains current.
FAQs
Frequently asked questions about Sourcegraph
How does Sourcegraph work?
Sourcegraph indexes repositories across a codebase and provides exact code search, natural-language Deep Search answers grounded in citations, and code intelligence for developers and AI agents.
What does Sourcegraph integrate with?
Sourcegraph supports GitHub and MCP-compatible coding agents. Its API compatibility includes Model Context Protocol (MCP), SCIP code intelligence, a streaming search API using SSE, and webhooks.
How much does Sourcegraph cost?
Sourcegraph's pricing model is Enterprise quote scaled by team size; outcome-based pricing per merged changeset for Agentic Batch Changes; free public-repository search.
What skills are needed to self-host Sourcegraph?
Sourcegraph's required skills are Repository and version-control administration, code search and navigation, site configuration; Kubernetes administration for Kubernetes-based self-hosting.
