Sourcegraph tour

Sourcegraph is a code intelligence platform for developers. It lets you search and explore your organization’s code on the web, with integrations into your existing tools. This tour walks through the most common ways developers use Sourcegraph every day.

Code Search

Sourcegraph code search is fast, works across all your repositories at any commit, and supports multiple search modes.

Search modes

Standard search
Structural search
Commit diff search

Standard search matches literal strings and regular expressions across all indexed files.

# Find all calls to a function
parseWithClaims

# Regex: find error-wrapping patterns in Go
lang:go fmt.Errorf\(.*%w

# Search in a specific repo
repo:my-org/my-service AuthenticationError

Use the filter bar in the search UI to narrow results by repository, language, file path, and more.

Structural search matches code patterns using Comby syntax, including nested expressions and multi-line code blocks. Unlike regex, structural search understands code structure.

# Match all arguments passed to fmt.Sprintf
fmt.Sprintf(...)

# Match try/catch blocks containing a specific call
try { :[body] } catch (:[e]) { :[handler] }

Structural search is particularly useful for finding all callsites of a function regardless of how arguments are formatted.

Diff search shows all commits that changed code matching your query, newest first.

# Find all commits that touched a function
type:diff myCrashingFunction

# Narrow to a time range
type:diff after:"3 months ago" AuthService

# Narrow by author
type:diff author:alice parseToken

Diff search is the fastest way to answer “what changed recently that could have caused this bug?”

Try it on open source

You can try Sourcegraph search on public repositories at sourcegraph.com/search. For example:

Search for repo:dgrijalva/jwt-go parsewithclaims to find all uses of the ParseWithClaims function

Code review: gain context on changed code

When reviewing a pull request, you often need to understand the full context of changed code—not just the lines in the diff.

Workflow

Install the browser extension

Install the Sourcegraph browser extension for Chrome, Firefox, or Safari. This adds code navigation directly to GitHub, GitLab, and Bitbucket pull requests.

Open a pull request

Visit a pull request in your code host. With the browser extension active, you’ll see hover tooltips appear when you mouse over code in the diff.

Use Go to definition

Hover over any symbol in the diff—added, unchanged, or deleted code. Click Go to definition to jump to the symbol’s implementation, even if it’s in a different repository.

Use Find references

Click Find references on a modified function to enumerate all call sites across your entire codebase. This helps you verify that every caller accounts for new behavior or assumptions introduced by the change.

To try this on an open-source pull request, install the browser extension and visit gorilla/websocket#342. Use Find references on the newly added AllowClientContextTakeover field to see all places it’s used.

Explore code: understand an unfamiliar codebase

As a new developer on a project, or when working in an unfamiliar part of the codebase, Sourcegraph’s code navigation helps you orient quickly.

Workflow

Search for the entry point

Search on Sourcegraph for the function or module you want to understand. Use the query bar on your organization’s Sourcegraph instance, or initiate a search from your editor or browser extension.

Navigate to the relevant result

Click through to the file. If the search returns many results, use the suggested filters below the search box or add filters like lang:go or repo:my-org/my-service to narrow them down.

Drill down with Go to definition

Click on any token and select Go to definition to navigate to its implementation. Continue drilling down depth-first until you’ve mapped the relevant call chain.

Share a link

If you have questions, use the blame information to identify who wrote the code. Copy the Sourcegraph URL for the relevant file and line, and send it with your question. Sourcegraph links include the exact revision, so the recipient sees exactly what you’re referring to.

Try this on the open-source node-jsonc-parser repository. Start at the getLocation function, follow Go to definition through visit, and continue until you reach createScanner.

Debug issues: trace what changed

When you receive a crash report with a stack trace, diff search helps you quickly identify what changed that might have caused the problem.

Workflow

Run a diff search on the crashing function

Take the function name from the top of the stack trace and run a diff search:

type:diff myCrashingFunctionName

Results show all commits that modified code matching that name, newest first.

Review recent changes

Scroll through the results. Each result shows the diff with context. Click on the commit message to view the full diff with code navigation enabled.

Drill into the change

Click on code in the commit diff to open the full file at the revision before or after the commit. Use Go to definition and Find references to understand how the change affected callers and behavior.

Narrow the search if needed

If there are too many results, add filters to focus on the relevant repositories and time window:

type:diff after:"last week" repo:my-org/my-service myCrashingFunctionName

Learn how a function should be called

When you need to add a new feature that uses an unfamiliar function or library, looking at real-world call sites is more reliable than reading documentation that may be outdated.

Workflow

Search for the function

Search on Sourcegraph for the name of the function you want to call.

Open a matching file

Find and click on a search result that looks like a relevant usage. If there are many results, narrow them using filters.

Find all references

Click on the function name in the file and select Find references to see all call sites across your codebase (and optionally, across public open-source repos).

Evaluate the examples

Click through to various call sites. Use the inline blame information to see who wrote each call site and when. More recent, non-test call sites by experienced team members are typically the best examples to follow.

Sourcegraph provides two tiers of code navigation:

Search-based code navigation

Precise code navigation

Code Navigation docs

Set up precise code navigation and explore all navigation features.

Cody AI

Cody is Sourcegraph’s AI code assistant. Because Cody has access to your entire codebase as context, its answers are grounded in your actual code rather than general programming knowledge.

Chat
Autocomplete
Commands

Ask Cody questions about your codebase in natural language:

“Where is the authentication middleware defined and how does it work?”
“Explain the retry logic in the payments service”
“What does this function return when the input is empty?”

Cody cites the specific files it used to answer, so you can verify and navigate to the source.

Cody overview

Learn about all Cody capabilities and supported clients.

Cody quickstart

Install Cody in your editor and get started.

Deep Search

Deep Search is an agentic AI tool that answers complex questions about your codebase. It explores your code autonomously, running multiple searches and synthesizing the results into a comprehensive answer. Unlike Cody chat, Deep Search is designed for open-ended investigative questions where the answer requires exploring many files and understanding broad patterns:

“How does authentication work across all our services?”
“Which repositories are still using the deprecated v1 API?”
“Walk me through how a request flows from the frontend to the database”

Use @-mentions to scope Deep Search to specific repositories, directories, files, or symbols for faster, more focused results.

Deep Search docs

Learn how Deep Search agents work and how to get the best results.

Code Insights

Code Insights turns any Sourcegraph search query into a time-series visualization. Use dashboards to track engineering metrics, migration progress, and codebase health over time. Example insights:

Track how many repositories still use an old version of a library (and watch the number decrease as migrations land)
Monitor the adoption of a new logging framework across all repos
Track the number of TODO comments or type: ignore annotations over time
Measure vulnerability remediation progress after a CVE disclosure

Code Insights docs

Learn how to create insights and build dashboards.

Batch Changes

Batch Changes automates large-scale code changes across many repositories. You write a changeset specification describing what to change, Sourcegraph applies it, and you manage the resulting pull requests from a single interface. Common use cases:

Update an API after a library major version bump
Apply a security patch to hundreds of repositories
Rename a configuration key across all services
Add required license headers to source files

Batch Changes docs

Learn how to create and manage large-scale code changes.

Get Started

Cody AI Assistant

Code Search

Platform Features

Code Search

Search modes

Try it on open source

Code review: gain context on changed code

Workflow

Explore code: understand an unfamiliar codebase

Workflow

Debug issues: trace what changed

Workflow

Learn how a function should be called

Workflow

Code Navigation

Code Navigation docs

Cody AI

Cody overview

Cody quickstart

Deep Search

Deep Search docs

Code Insights

Code Insights docs

Batch Changes

Batch Changes docs

Build docs developers (and LLMs) love

Get Started

Cody AI Assistant

Code Search

Platform Features

​Code Search

​Search modes

​Try it on open source

​Code review: gain context on changed code

​Workflow

​Explore code: understand an unfamiliar codebase

​Workflow

​Debug issues: trace what changed

​Workflow

​Learn how a function should be called

​Workflow

​Code Navigation

Code Navigation docs

​Cody AI

Cody overview

Cody quickstart

​Deep Search

Deep Search docs

​Code Insights

Code Insights docs

​Batch Changes

Batch Changes docs

Build docs developers (and LLMs) love

Code Search

Search modes

Try it on open source

Code review: gain context on changed code

Workflow

Explore code: understand an unfamiliar codebase

Workflow

Debug issues: trace what changed

Workflow

Learn how a function should be called

Workflow

Code Navigation

Cody AI

Deep Search

Code Insights

Batch Changes