#rag

Tokengram

tokengram

Token-saving memory layer for Claude Code — hybrid RAG search over your project (local; up to 70% on search/Read-heavy tasks, less on edit-heavy refactors).

CodeFlow-AI: Context-Aware Code Reviewer

SuyashZinjurke

Multi-agent context-aware code reviewer utilizing AST dependency graphs and RAG indexing.

Git Code Review

CodeReviewExtension

An Enterprise-ready code review governance workspace for code reviewers inside VS Code, with AI Coding Guard, Git context, Human-in-the-Loop findings, local-first RAG, cloud AI, Team collaboration foundations, SSO/RBAC, audit logs and AI governance.

Understand Tech

understand-tech

Chat with your Understand Tech AI assistants from inside VS Code — sidebar chat, file attachments, slash commands, and ask-about-selection.

BudAI

BudAI

Free — open-source models are the brains. Vibe-code with an AI that reads your folder and edits/runs your code (you approve every change, red/green diffs), build your own AI agents with tools, chat with your documents (RAG), web search, and multi-agent teams. 100% local.

OKAI

OKAI

OKAI v2 — Enterprise AI coding assistant with Accenture-grade UI, RAG knowledge base, and local LLM support.

Scaffold, visualize, and run AgentX agentic projects (LangGraph / CrewAI) from VS Code.

Jessie — AI Coding Agent

VijayArther

Jessie v3 — hosted BYOK AI coding agent. Code Review, Merge Review, tickets. Your API key stays on your device.

Persistent two-layer memory for GitHub Copilot. Auto-indexes your codebase and captures decisions, bug fixes and session knowledge — saving up to 90% of tokens in large projects.

Telos Code — Graphic Memory

telos

Turn your codebase into an interactive 2D force graph. AI-powered error diagnostics with visual path highlighting.

Analyze and repair Java static analysis findings with local LLMs, RAG context, traceable reports, and structured feedback.

memo — local memory for AI agents

jagoff

Registers memo — a local-first, long-term memory MCP server — in VS Code agent mode. MLX on Apple Silicon or CPU on Linux. Markdown on disk, no cloud, no API keys.