Engram

Acerca de

Engram acts as the "Missing Context" Engine for AI agents, bridging the gap between an LLM's understanding of code and the broader repository context. It leverages Git history to uncover hidden dependencies (Temporal Graph), extracts test intent to preserve required behaviors (Validation Graph), and provides a persistent knowledge store for architectural constraints (Knowledge Graph). This ensures AI agents can proactively identify and address potential issues, preventing silent breakages and accelerating development cycles, all while maintaining strict local-first privacy and zero telemetry.

Características Principales

• Knowledge Graph: A persistent store for LLMs to save and retrieve architectural decisions, edge cases, and project quirks.
• Local-First & Zero Telemetry: Ensures all processing happens locally with no data tracking for maximum privacy.
• Validation Graph: Automatically extracts specific test intent strings to provide behavioral guardrails for AI agents.
• 5 GitHub stars
• Temporal Graph: Mines Git history to identify frequently co-changed files, revealing hidden dependencies and blast radius.
• Adaptive Indexing Strategy: Optimizes performance for repositories of all sizes, from small projects to massive monorepos.

Casos de Uso

• Guiding AI agents to understand and preserve critical application behaviors defined by existing test suites.
• Preventing subtle regressions caused by hidden dependencies between seemingly unrelated files during AI-driven code changes.
• Capturing and sharing organizational memory and undocumented architectural constraints across AI agent sessions.