InkVault
Enables generation of optional daily digests (2-3 sentence summaries of each day) using a local model through Ollama.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@InkVaultFind that retry helper I saved"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
InkVault
Rescue your Pieces memory before it's gone, then search it from Claude, Codex, or any MCP client, and see it as a map of your year.
Pieces for Developers shut down on September 27, 2026. PiecesOS still runs in read-only mode on your computer, and your long-term memory is still inside it: every capture, session summary, chat and saved snippet. Once you uninstall PiecesOS, it's gone.
InkVault copies all of it into a single file on your machine, makes it searchable by your AI tools, and draws it as a private dashboard. Nothing is uploaded anywhere.
uvx --from git+https://github.com/sandoreclegane/inkvault inkvault rescueThat's the whole thing. It needs uv and PiecesOS running.
It takes a while. PiecesOS hands records over at about 8 per second, so a year of captures (~100,000) takes around 3-4 hours. Leave it running, or stop it anytime and run the same command later: it picks up where it left off.
What you get
The vault: everything PiecesOS will hand over (captures, summaries and their text, chats, snippets, tags, websites, people), saved as the raw records Pieces returned, so nothing is lost to interpretation. One SQLite file.
Search for your AI tools: an MCP server with
search_memories(keyword + meaning, merged),timeline("what was I doing in March?"),get_memory, andmemory_stats.Memory Atlas: a dashboard of your year. Active hours per day (hover a day for what it was about), when you work, how your projects connect, projects over time, top apps and sites. Every chart has a table view.
Daily digests (optional): a 2-3 sentence summary of each day, written by a local model through Ollama. Pieces stopped writing summaries; this picks up where it left off.
Related MCP server: Codex Native Memory
Connect your AI tools
Claude Code
claude mcp add inkvault --scope user -- uvx --from git+https://github.com/sandoreclegane/inkvault inkvault serveCodex (~/.codex/config.toml)
[mcp_servers.inkvault]
command = "uvx"
args = ["--from", "git+https://github.com/sandoreclegane/inkvault", "inkvault", "serve"]Hermes, Claude Desktop, and other MCP clients: run uvx --from git+https://github.com/sandoreclegane/inkvault inkvault serve over stdio.
Then ask: "What was I working on the week of March 10?", "Find that retry helper I saved", "When did I last look at the Stripe dashboard?"
Commands
Command | Does |
| Export, index, digest (if Ollama is running), build and open the dashboard |
| Copy everything out of PiecesOS. Safe to stop and re-run: it only adds what's new |
| Rebuild search |
| Daily digests with a local model (default |
| Rebuild and open the Memory Atlas |
| Run the MCP server |
| What's in your vault and where it lives |
Privacy
This is your screen history, your chats and your code. InkVault is built so that none of it leaves your machine:
Everything is stored in your app-data folder (
%LOCALAPPDATA%\InkVault,~/Library/Application Support/InkVault, or~/.local/share/inkvault), or whereverINKVAULT_HOMEpoints. Never inside the code folder.No telemetry, no accounts, no cloud.
Two things are downloaded, and nothing of yours is uploaded: the search model (~130 MB from Hugging Face, once) and the dashboard's chart library (from cdnjs, when you open the page).
Digests use Ollama on your own machine. Skip them with
--no-digest.Captured text was written by other people and apps. The MCP server tells your AI tool to treat it as data, not instructions.
Back up your vault. It is one file: vault.db in your InkVault folder. Everything else can be rebuilt from it
with inkvault index.
Good to know
Your projects on the dashboard are detected from recurring phrases in your session titles. To choose your own, create
themes.txtin your InkVault folder with lines likeWebsite = website|landing page|stripeand runinkvault dashboard.PiecesOS port: InkVault tries 39300, then 1000. If yours differs, set
INKVAULT_PIECES_PORTS.Pieces' own export tool: Pieces said they'd send one. Use it too; two copies are better than one. Support for importing its format into InkVault is planned.
Tested on Windows with PiecesOS 12.6.2 and a vault of 120,000 captures. macOS and Linux should work; reports are welcome.
Not affiliated with Pieces
InkVault is an independent, community tool for people moving on from Pieces. "Pieces" is a trademark of its owner.
MIT licensed. Made by T. Matthew Chase / Logos 7.
This server cannot be deployed
Maintenance
Related MCP Connectors
Search your AI chat history (ChatGPT, Claude, Codex) from any MCP client. Remote, private, read-only
Private persistent memory for Claude, ChatGPT & Gemini via MCP - semantic search, zero-code setup.
- mcpOAuthai.butlerbrain
Persistent memory for AI assistants. Save once; recall from Claude, ChatGPT, or any MCP client.
Personal knowledge MCP: capture bookmarks, notes & todos by chat; archive pages; search memory.
Related MCP Servers
- AlicenseAqualityAmaintenanceLocal-first AI PKM memory server for coding conversations. Imports Claude Code, Cursor, Codex CLI, Trae, and GitHub Copilot chats into notes, semantic search, tag graphs, Markdown exports, and MCP memory tools.7121 npm58Apache 2.0
- AlicenseBqualityCmaintenanceProvides local cross-session memory for Codex by importing transcripts, indexing in SQLite, and exposing search, context, and memory management through MCP tools.13MIT
- AlicenseNot gradedqualityBmaintenanceEnables AI tools to query a user's private, locally stored memories (notes, documents) with source citations, using the MCP protocol.34 npmMIT
- FlicenseNot gradedqualityBmaintenanceMCP server that indexes local Claude Code transcripts and provides keyword, phrase, semantic, and date-range search, plus full context/session recall.-