Skip to main content

aichivist

Clean up your files – like an archivist.

Your AI renames, sorts and de-duplicates your documents.
You approve every step, and everything can be undone.

tests Windows · macOS · Linux License: MIT

Get started · Why not just ask an AI? · What it can do

Why not just ask an AI?

Tell ChatGPT or Claude "tidy up my documents" and you usually get a quick job: folders by file type (PDFs, Images, Misc), the same cryptic names, duplicates left where they were – and sometimes a game or program that no longer starts because its files were moved. It happens all at once, with no plan to check and no way back. Next time, the AI invents a different system.

Asking an AI alone vs. an AI with aichivist

aichivist gives your AI what a professional archivist works with: your own filing rules, a plan you approve, names read from the document itself, duplicates found by content, programs that keep working, and a record of every step that can be undone. Tried in real use on 25,000 photos and 1,500 documents.

How it works

The AI reads, proposes a plan, you approve, everything is filed

  • Nothing happens without your OK. You see the plan first and can change or strike any line.
  • Nothing gets lost. Every step is recorded and can be undone; duplicates go to the trash only after you agree.
  • Your files stay yours. Normal folders, your own cloud, no account, no server.
  • Speaks your language. Messages in English or German; rule templates in English, German, French, Italian and Spanish.

What it can do

Reads scans, names in your style, finds duplicates, programs keep working, seals finished folders, retention periods, backup check, private stays private

Get started

Claude Desktop – the easiest way, no technical knowledge needed

  1. Download the .mcpb file from the latest release (no Claude Desktop yet? claude.ai/download).
  2. Double-click it and choose the folders Claude may organize.
  3. Ask Claude: "Tidy up my Downloads."
Other ways: Claude Code, Codex/Gemini CLI/Cursor, or without AI

Claude Code – type /plugin marketplace add luftphilipp/aichivist, then /plugin install aichivist@aichivist, then say "Set up aichivist." The setup installs the command-line tool if it is missing (it asks first).

Codex, Gemini CLI, Cursor – pipx install aichivist (or uv tool install aichivist), then copy templates/AGENTS.md to where your agent reads instructions.

Downloaded the ZIP? Unpack it and double-click Start (Mac).command or Start (Windows).bat (Linux: start.sh). A simple menu opens – look at your Downloads, check your backups, and see which AI app on your computer can tidy up. No installation needed, only Python 3 (a new Mac offers to install it by itself). The first time, your computer warns because the file comes from the internet:

  • Mac: click Done, open System Settings → Privacy & Security, scroll down and click Open Anyway next to "Start (Mac).command", then confirm.
  • Windows: click More info → Run anyway.

Just curious? See what's going on in your folders – no AI, nothing is changed:

pipx install aichivist
aichivist survey ~/Downloads

It shows how many files you have, duplicates that waste space, files with meaningless names, folders that slow down your cloud, program folders to leave alone, and whether your hard drive is healthy. aichivist storage storage.md checks whether your backups are safe and what your storage costs per year (it shows the table format on first run). aichivist catalog-page <folder> --open turns your catalogs into a searchable page, aichivist photos sorts photos by year, and aichivist watch ~/Downloads --install reminds you once 25 files have piled up.

See it in action (terminal demo)

aichivist demo

Built on real archive practice

Archives have organized documents reliably for centuries. aichivist brings their rules to your computer:

In an archive In aichivist
A filing plan says where everything belongs Your rules say which documents go where – the AI follows them
Nothing is thrown away without a decision Duplicates and old files are only removed after your OK
Every document is described in a catalog Each folder gets a list of what is in it, so you find things later
Finished collections are sealed A completed folder (e.g. last year's taxes) is locked, checked and searchable
Records are kept for a set time, then archived or destroyed Old records come up when their time is over – you choose
Copies are kept in separate places It checks the 3-2-1 rule: 3 copies, 2 kinds of storage, 1 away from home

Privacy

aichivist itself works only on your computer. The AI that reads your documents (e.g. Claude) receives the content of the documents it reads – so you decide which folders it may open, and can exclude private ones completely.

How this was built

aichivist was designed by Philipp Luft, a trained archivist and information professional (Fachmann Information & Dokumentation) studying Business AI, who brought the archival principles and the requirements. The code was written with Claude Code – every commit lists Claude as co-author.

More

Metadata

Release files for aichivist 0.1.0

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for aichivist 0.1.0
File Size Uploaded Provenance Actions
aichivist-0.1.0.tar.gz 113.3 kB With provenance Details for aichivist-0.1.0.tar.gz

Built distribution (wheel)

Table of built distributions (wheels) for aichivist 0.1.0
File Interpreter ABI Platform Actions
aichivist-0.1.0-py3-none-any.whl Python 3 none any Details for aichivist-0.1.0-py3-none-any.whl

Total release size: 210.9 kB

Release files / aichivist-0.1.0.tar.gz

Download URL aichivist-0.1.0.tar.gz
Size 113.3 kB
Tags Source
SHA-256 checksum
How to use checksums
afa51f94980dbe07d38a31c8e148067e31adc08cb5ffb5d6cf686c6e6426b160
BLAKE2b-256 checksum
How to use checksums
598835a55525c5644a206c4b926b2d90c10edf9f6910099138725616bc34f921
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/7.0.0 CPython/3.13.16

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Oct 27, 2026.

Transparency log

Release files / aichivist-0.1.0-py3-none-any.whl

Download URL aichivist-0.1.0-py3-none-any.whl
Size 97.6 kB
Tags Python 3
SHA-256 checksum
How to use checksums
7db1fa925d28e8513d68e44312c9dd72a4081df368c864fbdeeeb437c8c09105
BLAKE2b-256 checksum
How to use checksums
164eec4165c72ba0047afbf5f497eac8f1ce08234a36f5a62e4eed6871a31cea
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/7.0.0 CPython/3.13.16

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Oct 27, 2026.

Transparency log

Security for aichivist 0.1.0

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

Full provenance

All 2 files in this release have been uploaded with provenance metadata.

Select from 2 files

Attestation for aichivist-0.1.0.tar.gz

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Oct 27, 2026.

Transparency log

A note on package security

PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.

To learn more about protecting yourself from malicious actors, visit our package security documentation.

Release history Release notifications | RSS feed

This release

0.1.0 This release

2 release files Full provenance

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page