Plug our Context Engine into Claude Code, Cursor, Zed, and any MCP-compatible agent. Semantic code search that actually understands your codebase.
Install MCP001
Benchmark Results
We evaluated agents on a real-world task: take a natural language prompt and ship a complete PR. The dataset: 300 Elasticsearch PRs, each with 3 different prompts, for 900 attempts total.
Context architecture matters as much as model choice. A weaker model with great context can outperform a stronger model with poor context.
MCP-enabled runs consistently required fewer tool calls and conversation turns. Code that's more correct from the start means less time debugging.
300 Elasticsearch PRs · 900 attempts
Measured across 5 dimensions: correctness, completeness, best practices, code reuse, documentation
“If I have to work in Claude Code, I still always have the Augment Context Engine MCP installed.”
See it in action
Click to play demo
Demo: Context Engine MCP in action
002
Benefits
Our Context Engine retrieves the right information instantly. Your agent doesn't waste time grepping through files or exploring dead ends.
When agents don't have to grep and store irrelevant information in context, they use fewer tokens overall. Better signal, less noise.
Our indexer runs locally on your machine. When you make local changes, your next context query immediately reflects those changes.
Index code across multiple repositories, documentation sites, and internal wikis. Your agent sees everything relevant, not just one repo.
003
Supported Clients
If your favorite coding agent supports MCP, you can bring the Context Engine with you. We support multiple ways to set it up. Check our docs →
Pull in Git repos, documentation sites, and internal runbooks. Your agent searches everything, not just the repo you cloned.
Index entire orgs. GitHub, GitLab, Bitbucket.
Re-index on every push via CI/CD hooks.
Build custom indexers with our SDK.
005
FAQs
Get Started
Install Augment MCP to bring world-class codebase understanding to any MCP-compatible coding agent.