Sourcegraph Cody
Sourcegraph Cody is an AI coding assistant powered by cross-repository code intelligence. While other assistants see only the current project, Cody searches your entire codebase — across all repositories — to find relevant code, patterns, and usage examples. MCP tools add external system data to this already comprehensive codebase understanding.
Cross-Repository Intelligence Meets MCP
Cody's context engine is powered by Sourcegraph's code search index. Ask "how is authentication implemented?" and Cody finds the relevant code across your organization's repos — not just the one you have open. This cross-repo awareness makes Cody especially powerful for microservice architectures and monorepos.
MCP tools complement this by adding non-code context: deployment configurations, monitoring data, issue tracker statuses, and documentation stored outside your repos.
Features:
- Cross-repo search — find code patterns across all repositories
- Context engine — automatic relevance ranking for code snippets
- VS Code + JetBrains + Web — multiple editor support
- Agent mode — autonomous multi-step coding
- Multi-model — Claude, GPT, Gemini, and Mixtral
- Custom commands — reusable prompt workflows
- Enterprise — self-hosted and cloud deployments
- Open-source — Cody client is Apache 2.0
Setup
1. Create a Token
In Vinkius Cloud, go to your server → Connection Tokens → Create. Copy the URL.
2. Add MCP Server
In the Cody extension → Settings → MCP → add your Vinkius URL.
3. Search + Tools
Cody uses cross-repo intelligence and MCP tools together for complete context.
FAQ
Can Cody search across all my repositories? Yes. Powered by Sourcegraph's code search index, Cody finds relevant code across your entire organization.
Does Cody work in JetBrains? Yes. The extension is available for VS Code, JetBrains IDEs, and the web interface.
Can I self-host Cody? Yes. Sourcegraph offers self-hosted enterprise deployments.
Is Cody free? Free tier for individuals. Enterprise plans for teams.