Local-first AI project memory

Move your project between AI tools without rebuilding context

Memephant stores your project memory once — goals, key decisions, important files, and next steps — then hands a clean Context Passport to ChatGPT, Claude, Gemini, Grok, Codex, Cursor, or a local LLM. Switch AI tools without re-explaining your project every time.

Download free View on GitHub

Free during early access · Windows, macOS & Linux · No account needed

Works with ChatGPT, Claude, Gemini, Grok, Codex, Cursor, and local LLMs.

Every new AI chat starts from zero

AI tools forget your project between sessions. You re-explain your stack, paste old chat logs, and hope nothing important is missing. Memephant keeps one portable, local-first project memory so each AI session starts warm instead of blank.

How Memephant works

Store your project once — capture goals, key decisions, important files, and next steps, and scan a local folder to keep the memory grounded in reality.
Copy a Context Passport — a handoff shaped for the AI you're about to use, with known API keys, tokens, and local paths redacted before export where possible.
Paste, work, stay current — paste the AI's reply back into Memephant and review a diff before anything changes, so your memory carries to the next session and the next tool.

Explore Memephant

AI project memory — persistent, local-first memory for your projects across every AI tool.
Claude to ChatGPT handoff — move project context between Claude and ChatGPT without rebuilding it.
Local-first model — how your project memory stays on your device by default.
Download Memephant — get the desktop app for Windows, macOS, or Linux.