Engram indexes your repositories and gives AI assistants semantic search, knowledge graphs, and cross-repo context. Via MCP or transparent proxy—no code changes required.
# Ask your AI assistant about code it's never seen claude "How does the payment webhook handler validate signatures?" # Engram enriches the request with relevant context: engram: searched 3 repos, 847 files indexed engram: found 4 relevant symbols across 2 services PaymentWebhookController.cs:42 ValidateHmacSignature() WebhookMiddleware.cs:18 ISignatureValidator injection PaymentEvents.cs:67 WebhookReceived event handler docs/webhooks.md:1 Signature validation sequence # Your assistant responds with real answers, not hallucinated ones.
Vector embeddings + AST symbol extraction across 7 languages. Search by intent, not just text. Domain and component tags keep results precise.
Commits, work items, and code dependencies as a traversable graph. Multi-hop queries connect code to the business decisions behind it.
Fan-out queries across repo groups with rank fusion. Detects shared contracts, message queue boundaries, and API coupling between services.
25 MCP tools for Claude Code, Cursor, and others. Or use the transparent proxy to inject context with zero config—no agent cooperation needed.
Azure DevOps or GitHub. Git or TFVC. Webhook-driven indexing on every push.
AST parsing, vector embeddings, graph construction. Incremental after first run.
One line in your MCP config, or deploy the proxy. Your assistant gets codebase context immediately.
Index any VCS. Search from any AI tool.
Engram is deployed on your infrastructure. Custom installations available in-person or remote.
Get in touch to discuss your setup.