Mnemo
Persistent memory, semantic code search, cross-repo intelligence, and architecture awareness for AI coding assistants. One install gives your AI agent full project context across every chat session.
The Problem
Every time you start a new AI chat, your assistant forgets everything — what you decided, what you built, how your code is structured. You repeat context, re-explain architecture, and lose insights from previous conversations.
The Solution
Mnemo gives your AI agent a persistent brain. It remembers decisions, maps your codebase, searches code by meaning, and works across multiple repositories — automatically.
Capabilities
🧠 Persistent Memory
Your AI remembers across chat sessions:
- Architectural decisions and reasoning
- Bug fixes and their root causes
- Team preferences and conventions
- TODO items and follow-ups
- Production incidents and resolutions
🗺️ Code Intelligence
Your AI understands your codebase without reading every file:
- Full repo map with class structures, interfaces, and relationships
- Architecture pattern detection (Clean Architecture, CQRS, Hexagonal, Microservices)
- Dependency graph and impact analysis
- Code health reports (complexity hotspots, large files)
- Smart analyzer selection (Roslyn for C# when .NET SDK available, tree-sitter otherwise)
🔍 Semantic Code Search
Find code by meaning, not just filename:
- "Find token refresh logic" → finds
getToken(), acquireTokenSilent(), ClientCredentialTokenService
- "Show me error handling" → finds retry pipelines, DelegatingHandlers, catch blocks
- "Database access code" → finds CosmosDbService, repositories, connection code
- Powered by ChromaDB (auto-installed, zero config)
🔗 Multi-Repo Workspace
Search across your entire platform:
- Link sibling repos with one command
- Cross-repo semantic search ("find auth code across all services")
- Cross-repo impact analysis ("what breaks in other repos if I change this?")
- Auto-discover and link all repos in a directory
📚 Knowledge Base
Your team docs become searchable by the AI:
- Architecture docs, runbooks, coding standards, gotchas
- Searched by meaning, not just keywords
- Add markdown files to
.mnemo/knowledge/
🌐 API Discovery
Your AI knows every endpoint:
- Parses OpenAPI/Swagger specs automatically
- Detects controller annotations (ASP.NET, Express, etc.)
- Semantic search across all discovered endpoints
👥 Team Intelligence
Know who owns what:
- Code ownership from git history (excludes merge commits)
- Team expertise map — who knows which service
- "Who last modified this file?"
📋 Task-Aware Context
Your AI focuses on what matters:
- Set active task: "I'm working on JIRA-456"
- AI automatically retrieves relevant code for your task
- Task completion tracking
🐛 Error & Incident Memory
Never debug the same issue twice:
- Store error → cause → fix mappings
- "Have we seen this NullReferenceException before?" → finds the stored fix
- Record production incidents with root cause and prevention
📝 Code Review Memory
Learn from past reviews:
- Store review feedback and outcomes
- Track rejected suggestions so AI doesn't repeat them
- Reference past review agreements
🚀 Auto-Remember
The AI saves important findings automatically:
- Code changes that affect behavior
- Bug fixes and their solutions
- Architecture decisions made during conversation
- Non-obvious insights about the codebase
Getting Started
- Install this extension
- Open a project folder
- Click "Yes" when prompted to initialize Mnemo
- Start chatting — your AI now has full project context
No terminal commands. No config files. No setup friction.
What to Ask Your AI
You don't need special syntax — just ask naturally:
| What you want |
What to say |
| Project overview |
"What do you know about this project?" |
| Find code |
"Show me the AuthorizationService methods" |
| Follow patterns |
"Show me existing handlers I can follow" |
| Architecture |
"What's the architecture of this project?" |
| APIs |
"What API endpoints exist?" |
| Save context |
"Remember we chose Redis for caching" |
| Impact analysis |
"What breaks if I change AuthService?" |
| Cross-repo |
"Find authentication code across all services" |
| Code health |
"What's the code health?" |
| Tests |
"What tests cover this file?" |
| Team |
"Who knows about the payment service?" |
| Errors |
"Have we seen this error before?" |
| Task tracking |
"I'm working on JIRA-456" |
| Incidents |
"Record this outage" |
Works With
- Amazon Q — Full MCP integration
- Cursor — Full MCP integration
- Claude Code — Full MCP integration
- GitHub Copilot — MCP support
- Kiro — MCP support
- Any MCP-compatible client
Extension Commands
| Command |
What it does |
| Mnemo: Initialize Workspace |
Set up Mnemo in your current project |
| Mnemo: Show Status |
Run diagnostics, check MCP server health |
| Mnemo: Refresh Index |
Rebuild code index after major changes |
| Mnemo: Check Installation |
Verify Mnemo is available |
Supported Languages
- C# (.cs) — Enhanced with Roslyn when .NET SDK available
- Python (.py)
- JavaScript (.js, .jsx)
- TypeScript (.ts, .tsx)
- Go (.go)
How It Works
You install extension → Extension downloads Mnemo → Mnemo indexes your code
→ AI client connects via MCP → Every chat has full project context
No Python required on your machine. No PATH configuration. The extension handles everything.
Requirements
- VS Code 1.92+
- Git (for team graph and change detection)
- An AI client with MCP support
Links
