Synapse automatically captures your AI coding sessions across every tool you use. When you go idle, it distills the valuable parts — decisions, architecture, learnings — into a curated knowledge layer your next session opens with.
Hundreds of sessions full of decisions, architecture, and debugging insights
Trapped in chat logs you’ll never scroll back to
Every morning, you re-explain your own project to your own tools
The knowledge is there — it just never gets extracted
A background daemon silently records your AI coding sessions across every tool you use
An LLM extracts decisions, architecture, and learnings — the signal from the noise
Next session opens with a brief from the workspace — your AI tools have context from every past conversation, no re-briefing needed
Sign up, start capturing. Your sessions become knowledge automatically.
$ npm install -g synapsesync \
&& synapsesync wizard \
&& synapsesync capture startWorks with any AI coding tool that uses Anthropic, OpenAI, or Google APIs.
A background daemon captures every AI session — across tools, through the proxy or per-tool file-watcher. No plugins, no prompts, no manual saving.
An LLM reads your sessions and extracts decisions, architecture notes, and learnings. Signal from noise, automatically.
Knowledge captured in one AI tool is available in every other. From your laptop to your desktop. Across your whole team.
Search by meaning, not keywords. “Auth flow” finds documents about login, session tokens, and OAuth.
Set a passphrase and content is encrypted before it leaves your machine. The server never sees plaintext.
Two paths. (1) A lightweight background daemon watches where AI tools store session data on disk and parses new files into a standard format — works for the tools we have file-watcher adapters for. (2) A built-in TLS-MITM forward proxy intercepts traffic to the Anthropic, OpenAI, and Google APIs — works for any tool that honors HTTPS_PROXY, with no per-tool adapter needed. No plugins, no manual saving.
When a session goes idle, Synapse sends the transcript to an LLM (your choice of Anthropic, OpenAI, or Google) that extracts decisions, architecture notes, and learnings. Trivial exchanges are skipped. Only the valuable stuff makes it into your workspace.
Any tool that talks to the Anthropic, OpenAI, or Google API — Cline, Cursor, Codex, Claude Code, Gemini, OpenCode, Aider, custom scripts, and more — is captured via the built-in proxy with no per-tool setup. File-watcher adapters add an extra capture path for Claude Code (via hooks), Cline, Cursor, Codex, Gemini, GitHub Copilot CLI, and Roo Code. The knowledge workspace is accessible from any MCP-capable tool, including ChatGPT, Windsurf, and VS Code.
Yes. You can enable end-to-end encryption with a passphrase — content is encrypted on your machine before it ever hits the server. We cannot read your files. Captured sessions are stored locally by default and only sync to the cloud if you have an API key configured.
Free tier: up to 50 projects with all capture features, MCP access, and 1-on-1 email invites. Plus at $5.99/mo adds AI-generated project context, link-based sharing, unlimited team members per project, and version history. You provide your own LLM API key for distillation.
Capture sessions, distill knowledge, remember everything. Free to start.