Sourcegraph Cody
Cody is an AI coding assistant by Sourcegraph that uses advanced codebase context and the latest LLMs to help developers understand, write, and fix code faster.
At a Glance
About Sourcegraph Cody
Cody is an AI coding assistant built by Sourcegraph that leverages deep codebase context — including semantic search across local and remote repositories — to help developers understand, write, and fix code. It is available as a VS Code extension, a JetBrains plugin, and on the web, and supports multiple large language models including Anthropic Claude Sonnet 4 and OpenAI GPT-4o.
What It Is
Cody sits in the AI coding assistant category, operating directly inside popular IDEs and on the web. Its core job is to give developers and AI agents accurate, full-codebase context so that code suggestions, chat answers, and automated edits reflect the actual state of a project — not just the currently open file. Sourcegraph describes Cody as an "AI coding agent with the best codebase understanding," positioning it against tools that only see a narrow slice of a repository.
Core Capabilities
Cody ships with several distinct interaction modes:
- Chat: Ask questions about the codebase; Cody uses semantic search to retrieve relevant files and answers with cited context. Enterprise users can add remote repositories as context.
- Autocomplete: Single-line and multi-line suggestions as you type, reducing the need to look up function and variable names.
- Inline Edit: Request fixes or refactors from anywhere in a file without leaving the editor.
- Prompts: Quick, customizable prompts for common tasks such as "Document code," "Explain code," and "Generate Unit Tests."
- Swappable LLMs: Supports Anthropic Claude Sonnet 4, OpenAI GPT-4o, Mixtral, Gemini 1.5, and more.
Editor and Platform Support
Cody is available for:
- VS Code via the VS Code Marketplace (extension ID:
sourcegraph.cody-ai) - JetBrains IDEs via the JetBrains Marketplace (plugin ID: 9682)
- Web at sourcegraph.com/cody/chat
The VS Code extension had version vscode-v1.116.0 ("Cody for VS Code 1.116.0") published on July 30, 2025, indicating active ongoing development.
Codebase Context Architecture
What differentiates Cody from simpler autocomplete tools is its integration with Sourcegraph's broader code intelligence platform. Cody can pull context from across thousands of repositories using SCIP-powered code navigation, symbol search, and semantic search. For enterprise deployments, this means agents and developers get answers grounded in the full codebase rather than a single file or project. Sourcegraph's MCP server extends this context to external AI agents such as Claude Code, Cursor, Codex, and Amp.
Repository and Open-Source Status
The Cody codebase was originally developed as an open-source project under the Apache License 2.0. In 2025, Sourcegraph migrated the main repository to a private repo; the public snapshot at sourcegraph/cody-public-snapshot preserves the code as it was just before that migration. The last commit under the Apache License is documented in the snapshot repository. The public snapshot is archived and shows 3,798 stars and 484 forks as of the data collected.
Update: Cody for VS Code 1.116.0
The latest release in the public snapshot is vscode-v1.116.0, published July 30, 2025. The repository's last push was August 1, 2025, confirming the product remains under active development even as the primary codebase moved to a private repository. Sourcegraph continues to ship Cody as part of its broader platform alongside Deep Search, Code Search, Batch Changes, and the MCP server.
Community Discussions
Be the first to start a conversation about Sourcegraph Cody
Share your experience with Sourcegraph Cody, ask questions, or help others learn from your insights.
Pricing
Free
Free tier for individual developers with access to Claude Sonnet 4 and other models, subject to per-user rate limits.
- Access to Anthropic Claude Sonnet 4
- Code autocomplete
- Chat with codebase context
- Inline edit
- VS Code and JetBrains extensions
Pro
For individual developers who need higher usage limits and additional model access.
- Higher usage limits than Free
- Access to multiple LLMs
- Code autocomplete
- Chat with codebase context
- Inline edit
- VS Code and JetBrains extensions
Enterprise
For enterprise teams needing dedicated instances, audit logs, RBAC, SSO, self-hosting, and bring-your-own-model support.
- Single-tenant cloud or self-hosted deployment
- Bring-your-own-model
- Audit logs
- SSO (SAML, OpenID Connect, OAuth)
- SCIM provisioning
- RBAC
- Remote repository context
- MCP server access
- 24x5 support
- Customer success manager
- SOC2 Type II + ISO27001 compliance
- Zero data retention
Capabilities
Key Features
- AI chat with codebase context
- Inline code autocomplete
- Inline edit and refactor
- Customizable prompts (Document, Explain, Generate Unit Tests)
- Swappable LLMs (Claude Sonnet 4, GPT-4o, Gemini, Mixtral)
- Semantic search across local and remote repositories
- @-mention files for targeted context
- VS Code extension
- JetBrains plugin
- Web chat interface
- Enterprise remote repository context
- Bring-your-own-model support (Enterprise)
Integrations
Demo Video

