rawmem MCP server

rawmem

English | 简体中文

<!-- mcp-name: io.github.Liyuan1992/rawmem -->

rawmem is a tiny local-first evidence ledger for AI and human workflows.

The idea is deliberately small:

Store now, understand later.

AI memory systems often start by summarizing. rawmem starts one layer lower: append-only raw evidence. A chat turn, a 2px UI tweak, a bug fix, a command run, a git snapshot, a browser clip, or a coding-agent completion can all become the same kind of local event.

The hard boundary is simple: capture is not recall. The ledger is a source layer. Review, filtering, MemoryDSL conversion, preference mining, and project rules can be derived later.

Why This Exists

Small facts look noisy alone. Repeated across time and projects, they become patterns:

• "This icon is too small; make it 2px bigger."
• "The AI forgot to update the desktop package again."
• "This project prefers append-only bug records."
• "The same class of boundary bug appeared in three repos."

Disk is cheap. Missing history is expensive.

Install

New to programming? Choose the Windows app (EXE). An EXE is a Windows program you open by double-clicking. Running it requires no Python installation or terminal commands.

Choose one approach; you do not need both. There is currently no downloadable release of the new EXE preview. A developer must build it from source first. Users unfamiliar with programming can wait for an EXE release instead of following the Python commands below.

Windows app (EXE preview)

Once you have a built rawmem.exe, double-click it, choose a storage folder, decide whether to start at logon, and click to begin recording. Browser saving requires the optional extension. See Windows recorder for setup and build instructions. This preview is unsigned and the extension still requires manual developer-mode installation.

Python CLI (for developers)

EXE users can skip this section. With Python 3.10+ and Git installed, run these commands in PowerShell:

git clone https://github.com/Liyuan1992/rawmem.git
cd rawmem
python -m pip install --user -e .

The core runtime has no third-party dependencies. DeepSeek Harness compressed session capture and the MCP server are optional extras:

python -m pip install --user -e ".[deepseek-harness,mcp]"

To install the published v0.7.1 wheel without cloning the repository:

python -m pip install "rawmem[deepseek-harness,mcp] @ https://github.com/Liyuan1992/rawmem/releases/download/v0.7.1/rawmem-0.7.1-py3-none-any.whl"

Make sure the user Scripts directory (e.g. %APPDATA%\Python\Python312\Scripts) is on PATH so rawmem works from any directory.

One-Time Setup

rawmem setup --global --install-startup --dry-run  # inspect first; writes nothing
rawmem setup --global --yes           # global config + git hooks for all repos
rawmem setup --install-startup --yes  # run the daemon headless at every logon
rawmem setup --start-daemon           # start it right now
rawmem config --show-browser-token    # paste into the browser extension options
rawmem doctor                         # verify the complete installation

Shortened. The full README is on GitHub.