Persistent memory for AI coding agents
Your AI agent logs decisions after every task and searches project history before starting new work — via MCP tools. Git push auto-capture is the built-in fallback.
How it works
Your AI agent gets persistent project memory through four capabilities.
Agent logs reasoning
Your AI agent calls log_context after every task — recording decisions, risks, and rationale. The WHY behind code is captured automatically.
Agent searches context
Before starting work, your agent calls search_context to ask "what did we decide about auth?" and gets a grounded, citation-backed answer.
MCP-native integration
Works with GitHub Copilot, Cursor, Claude, and any MCP-compatible agent. Five built-in tools — zero custom glue code required.
Auto-capture fallback
Not using an AI agent? Every git push still auto-extracts context via Nova Pro. Your project brain grows either way.
Up and running in 3 steps
Works with any git repo. Your AI agent gets memory in minutes.
Install & initialize
Download the VSIX, install it in VS Code, and click the ⚡ FlowSync status bar button. Initialize your project — you get a Project ID and Token.
Connect your AI agent
Your AI agent now automatically has 5 new tools: log_context to record decisions, search_context to query history, and 3 more.
Code with memory
Your agent logs WHY after every task and searches context before starting new work. Git pushes are auto-captured as a fallback. Open the dashboard for team visibility.
