Managed Knowledge Indexing

Sypha's Managed Indexing engine facilitates high-context semantic search across your entire codebase using cloud-hosted vector embeddings. When initialized, Sypha generates a mathematical map of your project, allowing the AI to deliver significantly more relevant and context-aware responses by "understanding" the relationships between different modules.

Strategic Capabilities

• Semantic Discovery: Perform conceptual searches across your entire repository anatomy.
• High-Fidelity Responses: Experience AI interactions that are deeply grounded in your project's technical reality.
• VCS-Aware Analysis: Intelligently tracks logic differences between your base branch and active feature branches.
• Collaborative Knowledge: Shared indices are available for team-based and enterprise-level environments.

Integration Prerequisites

• Git Repository Architecture: Indexing is restricted to Git-initialized root directories.
• Developer Credit Balance: Managed indexing utilizes cloud-resident compute and is debited from your active Sypha credits.

Configuration & Governance

Managed Indexing is active by default to ensure an optimal experience. To modify this behavior, initialize a .sypha/config.json file in your repository root:

{
  "project": {
    "managedIndexingEnabled": false
  }
}

Architectural Execution

• Base Branch Sync: The primary codebase is indexed in its entirety to establish a technical baseline.
• Feature Layering: Only the logical differences (diffs) from the base branch are incrementally indexed during active development.
• Vector Storage: Processed embeddings are persisted in Sypha Cloud. Note: Your raw source code is never stored; only the mathematical vector representations are retained.
• Lifecycle Management: Indices are persisted for 7 days and are automatically decommissioned if no updates occur within that window.

Strategic Benefits

Managed Indexing is the preferred solution for massive codebases where locating technical context is difficult, providing engineering teams with a shared, persistent semantic understanding of their infrastructure.