Text and Markdown documents
Plain text, Markdown, CSV and other UTF-8 files are always read. Attach them to a conversation, or add them to a project from the Documents page.
For students and researchers
Add your texts and reading notes to a project so the assistant starts from them, and each passage stays tied to its file. This path is partial: PDF and Office files are not read yet in the published app.
In the product todayPartly covered
Proofs last checked on 7 Oct 2026
“I have forty papers and the assistant forgets the ones it read.”
What you learned from each source is lost between conversations, so the reading starts again every time.
“I cannot tell which paper a sentence came from.”
A summary without its source cannot go into a thesis. You reopen the files one by one to check.
“I paste the same excerpts at the start of every conversation.”
Bringing the assistant up to speed takes longer than the question you wanted to ask.
“I do not remember why I dropped that study.”
The reason you kept or excluded a source lives in your head or in a conversation you can no longer find, and your supervisor will ask.
“Some of my material is unpublished or confidential.”
You need to know what stays on your computer and what is sent to the company running the assistant.
Plain text, Markdown, CSV and other UTF-8 files are always read. Attach them to a conversation, or add them to a project from the Documents page.
Each document is cut into passages that keep their position in the file, and their page or sheet when the format has pages. The original stays on your disk and opens from the Documents page.
Write what you learned from a source as a note, with a type and tags. The assistant can search your notes by meaning, and you can confirm a note or invalidate it with a reason when a source contradicts it.
Record the choice with its reasons and the alternatives you set aside, as a Decision. It is there when your supervisor asks, and the assistant can find it by meaning too.
The reading code exists for all four, but it sits behind build options that the published app does not turn on. Today a PDF or Word file is refused, and an Excel or PowerPoint file is kept without its text.
Passages are indexed when you upload a document, and the search exists in the data layer. No screen and no assistant tool calls it yet.
About twenty papers, your annotations in Markdown, one research question. The same month, without and with Project Orchestrator.
YouAt the start of each session you paste excerpts of your annotated papers to bring the assistant up to speed.
AssistantIt answers with a fluent summary, and nothing says which paper a sentence comes from.
YouYou reopen the files to check each claim, one by one.
YouWeeks later you no longer remember why you dropped one study, so you read it again.
YouYou create a project for the review and add your annotations as text or Markdown files. Each original is kept on your disk.
Project OrchestratorEach file is cut into passages that remember where they sit in it. The Documents page lists the files and opens the originals.
YouIn the chat you attach the file you are discussing and ask for quotes, not only a summary. You check each quote in the original.
YouYou write the lesson as a note tagged with the source, and record “dropped because of the sample” as a Decision.
AssistantIn the next session it can search your notes and decisions by meaning instead of starting from zero.
You still check every quote yourself. Project Orchestrator keeps the trail between your files, notes and choices; it does not certify what the assistant says.
You say “my papers” and “my reading notes”. Project Orchestrator says Document and Note.
“a thesis, a literature review”
Project
A body of work with its own plans, notes and decisions.
“my research question”
Objective
What the work is heading toward.
“my PDFs and papers”
Document
Spreadsheets, decks and files attached to your work.
“the paper I give it to read”
Attachment
A file added to a conversation or a project; the assistant reads its text when the format is supported.
“my reading notes”
Note
What was learned: guidelines, gotchas, patterns, tips.
“why I kept or dropped a source”
Decision
A choice that was made, with its rationale and alternatives.
“the AI that reads for me”
Assistants
A conversation with an AI that works on your project; Today shows which ones are running.
“finding what I wrote by its idea”
Semantic search
Finding notes and decisions by what they mean, not by their exact words.
Words this page leaves out : embedding, chunk, vector, knowledge graph
A document is cut at sentence boundaries into passages of about 1,500 bytes, contiguous and without gaps. Each keeps its start and end in the file and its order, plus its page when the format is paginated; plain text has no pages.
The original bytes are kept in a folder of the application on your computer, filed under their SHA-256, so the same file added twice is stored once. A single file is capped at 64 MiB.
The default model that turns passages and notes into searchable form runs locally, in the app, and handles French, English and about a hundred other languages. It is downloaded at the first setup.
An attached document reaches the assistant as text under its file name, capped at 60,000 bytes per document: the beginning is sent and the assistant is told the rest was cut.
The assistant’s note and decision tools include a search by meaning (action search_semantic). Documents are not covered by it yet.
The PDF, Word, Excel and PowerPoint readers are Cargo options that are off by default, and the release workflow builds without them. Without the option, a PDF or Word upload is refused with the message that the support is not compiled in.
Nothing here is hidden: these are the things you might assume and that are not true yet.
PDF, Word, Excel and PowerPoint are not read in the published app: a PDF or Word file is refused, an Excel or PowerPoint file is stored without text. Reading them means building the server yourself with the options pdf, docx, xlsx and pptx.
A PDF made of page images, such as a scan, has no text to read, even with the option turned on. There is no text recognition (OCR).
Only the first 60,000 bytes of a document reach a conversation. A long text, a whole thesis for instance, is read from the beginning only; split it or attach the part you are discussing. The chat takes at most 10 attachments per message.
The assistant gets no page numbers. Pages are kept with the passages and reachable through the server’s API, but no screen lists passages yet.
No search by meaning across documents. It exists for notes and decisions only.
Nothing checks that a quote really appears in its source. The matching code exists, and no screen or tool calls it yet.
Text files must be UTF-8. A file in an older encoding or in UTF-16 is stored without text.
No measurement of very large collections has been published, and the Documents page lists a document only if it belongs to a project.
Download it for your system. The first launch needs Docker Desktop and takes a few minutes; the setup also lets you pick the assistant for the chat.
DownloadOne project per thesis or review keeps its documents, notes and decisions together, and gives the assistant one place to look.
On the Documents page, pick the project and upload .txt, .md or .csv files: your annotations, or the text of a paper exported to a text file. A PDF will be refused for now.
Attach one file to the chat, ask for quotes with your question, then use Open original on the Documents page to check each quote.
Write the lesson of each reading as a note tagged with the source, and record every kept or dropped source as a Decision.
How it worksNot in the published app today. A PDF or Word file is refused with a message that the support is not compiled in, and an Excel or PowerPoint file is kept without its text. Text, Markdown and CSV work. The reading code exists behind build options, and it is not decided yet whether the standard download will turn them on. Until then, export the text of a paper to a .txt or .md file and add that.
The original files, the passages and the notes are stored on your machine, and the indexing model runs there too; there is no hosted version. But whatever the assistant reads in a conversation, the attached text and the notes it looks up, is sent to the company that runs the model you chose, Claude by default. For confidential or embargoed material, read that provider’s terms first. Project Orchestrator makes no compliance claim.
It receives the text under the file name, without page numbers, so it can name the file but not reliably the page. Pages are stored with the passages for formats that have them, but they are not given to the assistant yet. Ask for quotes and check them in the original: nothing in the app checks a quote for you today.
Not yet. It can search your notes and decisions by meaning. For documents, it reads the ones you attach to the conversation. That is why the habit that works today is to write what you learned from each source as a note.
The default indexing model is multilingual, so notes in French and in English are searched together. The assistant’s own languages are those of the model you chose.
No. There is no integration with reference managers and no citation formatting. A BibTeX file is plain text, so it can be attached like any text file, but nothing parses it.
The app is free and open source (MIT). The assistant you connect has its own cost. You do not write code, but the installation needs Docker Desktop, and the chat needs an assistant such as Claude Code that the setup helps you install.
Download
One-click install
There is no Project Orchestrator account to create.
Free · MIT license · Bring your own AISee what it costs
Version 0.0.19 · 13 Nov 2026
The app installs on a computer. Pick the one you use.
Prefer a terminal?
Homebrew · Mac and Linux. Installs orchestrator, orch and mcp_server.
brew install this-rs/tap/project-orchestratorDocker Desktop
Runs the database and the search engine the app relies on. The app starts them for you. Get Docker Desktop
Claude Code
The default engine for chat. The AI itself is not included: with Claude Code, you use your own Claude account, and the app can install it for you. Other engines (Codex, an OpenAI-compatible model, an ACP agent) are available from Settings — newer, in alpha. Get Claude Code
At first launch
A setup wizard in four short parts checks Docker, installs Claude Code if it is missing and downloads the search model.
Runs the two services that hold the project memory (Neo4j and Meilisearch).
Open the downloaded .dmg, drag Docker into Applications, then open Docker once and wait for the whale icon to stop moving.
Or in a terminal
brew install --cask dockerCheck that it works
Each line should print a version number. If you see « command not found », open a new terminal window and try again.
docker --versiondocker compose versionDocker must be running (open it once) before you start Project Orchestrator. Installation page of the publisher
The assistant Project Orchestrator works with. It uses your own Claude account.
Paste the command in a terminal. It installs everything it needs and asks for your password only if required.
Command
curl -fsSL https://claude.ai/install.sh | bashCheck that it works
Each line should print a version number. If you see « command not found », open a new terminal window and try again.
claude --versionRun claude once to sign in. Installation page of the publisher
Free · MIT license · Bring your own AI See what it costs
Try it: start a task in any window
Download
Pick the file for your system above. There is no account to create.
Open
Open the file. A setup wizard checks Docker, installs Claude Code if it is missing and downloads the search model.
Connect your AI
Chat runs on Claude Code by default; other engines are available from Settings (alpha). If it asks you to sign in, type this once in a terminal.
claudemacOS, Apple Silicon
The app for Macs with an M1 chip or later.
Project.Orchestrator_0.0.19_aarch64.dmg · 87.1 MB
macOS, Intel
The app for Macs with an Intel chip.
Project.Orchestrator_0.0.19_x64.dmg · 89.8 MB
macOS, Intel (macOS 10.15 to 13.3)
For Intel Macs on macOS 10.15 to 13.3. Without local embeddings: use an HTTP provider such as Ollama.
Project.Orchestrator_0.0.19_x64-legacy.dmg · 47.7 MB
Linux, AppImage
One file that runs on most Linux distributions (x86_64).
Project.Orchestrator_0.0.19_amd64.AppImage · 163 MB
Debian and Ubuntu, x86_64
Package to install with dpkg or apt.
Project.Orchestrator_0.0.19_amd64.deb · 98.1 MB
Debian and Ubuntu, arm64
Package to install with dpkg or apt, for ARM64 machines.
Project.Orchestrator_0.0.19_arm64.deb · 94.4 MB
Fedora and RHEL, x86_64
Package to install with dnf or rpm.
project-orchestrator-0.0.19-1.x86_64.rpm · 61.5 MB
Fedora and RHEL, arm64
Package to install with dnf or rpm, for ARM64 machines.
project-orchestrator-0.0.19-1.aarch64.rpm · 56.7 MB
Homebrew
Mac and Linux. Installs orchestrator, orch and mcp_server.
brew install this-rs/tap/project-orchestratorInstall script
Mac and Linux. Latest release into ~/.local/bin. Add --no-frontend for the light variant.
curl -fsSL https://raw.githubusercontent.com/this-rs/project-orchestrator/main/install.sh | shPowerShell
Windows. Installs into %LOCALAPPDATA%\orchestrator\bin.
irm https://raw.githubusercontent.com/this-rs/project-orchestrator/main/install.ps1 | iexDocker
Image for amd64 and arm64, with the web interface.
docker pull ghcr.io/this-rs/project-orchestrator:latest# API only, no dashboard: a version tag with the -api suffix# docker pull ghcr.io/this-rs/project-orchestrator:0.0.19-apiorchestrator, orch and mcp_server, with the web interface included. Lighter versions without it are on the release page.
macOS, Apple Silicon
Programs without the app window: orchestrator, orch, mcp_server
orchestrator-full-0.0.19-macos-arm64.tar.gz · 81.9 MB
macOS, Intel
Programs without the app window: orchestrator, orch, mcp_server
orchestrator-full-0.0.19-macos-x86_64.tar.gz · 82.8 MB
Linux, x86_64
Programs without the app window: orchestrator, orch, mcp_server
orchestrator-full-0.0.19-linux-x86_64.tar.gz · 91.1 MB
Linux, arm64
Programs without the app window: orchestrator, orch, mcp_server
orchestrator-full-0.0.19-linux-arm64.tar.gz · 85.6 MB
Windows, x86_64
Programs without the app window: orchestrator.exe, orch.exe, mcp_server.exe
orchestrator-full-0.0.19-windows-x86_64.zip · 92.8 MB
The desktop app starts Neo4j and Meilisearch in Docker for you. With the command line, one compose command does it: follow the quickstart.
More for developersWhat you should see. A list of checks. On the very first launch the Docker lines say “Setup pending”.
What to do. Wait for the setup to open: the start screen only checks what is installed.
What you should see. A green line says “Docker Desktop is running” under two cards, “Docker (recommended)” selected.
What to do. Keep “Docker (recommended)”. The app looks for Docker Desktop. When it says “Docker Desktop is running”, press Next. If it says Docker is missing or not running, use the button next to the message, then wait: the app checks again by itself. The services use the local ports 7474, 7687, 7700 and 4222, and the app itself uses 6600.
What you should see. Three choices (No authentication, Password, OIDC / OAuth), the first one selected.
What to do. On a computer only you use, keep “No authentication” and press Next. On a shared or public network, choose “Password” and set an email and a password: without sign-in, the app’s server accepts connections from your whole network.
What you should see. Claude Code CLI shows a green check once installed (until then: “Claude Code CLI is required” and an install button), and the search model is marked ready.
What to do. If “Claude Code CLI is required” is shown, press “Install Claude Code CLI”. Then let the app download the search model when it offers to; the size is shown on screen. Next unlocks when both are ready. If you have not signed in to Claude yet, see “If it does not work” below.
What you should see. A green “Configuration saved!” message and a “Restart Application” button.
What to do. Press “Generate Config & Save”. When “Configuration saved!” appears, press “Restart Application”. A switch on this screen controls whether the app checks for updates at start-up.
What you should see. A checklist (Docker Desktop, Neo4j, Meilisearch, NATS, Backend server) turning green one line at a time.
What to do. On this second start the app downloads the three images it runs on, then starts them. Later starts reuse the images and are quicker. When every line is green, the main window opens. If a line says “Taking longer than expected”, wait. If a “Continue anyway” button appears, see “If it does not work”.
macOS 11 or later on Apple Silicon (M1 and later). On Intel: macOS 13.4 or later, or macOS 10.15 to 13.3 with the older-macOS file, which has no local embeddings.
Download Docker Desktop from Docker’s page, open the file, drag Docker to Applications, then open it and wait until it says it is running.
Get Docker DesktopDocker Desktop is open and reports that it is running. It is Docker’s own screen, so it is not reproduced here.
Open the .dmg, drag Project Orchestrator to Applications, then open it from there. If macOS says it cannot check the app, we could not verify how the build is signed: allow it in System Settings, Privacy & Security, with “Open Anyway”.
The three problems you are most likely to meet, with what the app shows.
What you see. The app says “Docker Desktop is not running” or “Docker Desktop is required”.
What to do.
What you see. Neo4j or Meilisearch stays “Unhealthy”, or the app says it failed to start services. The services need the local ports 7474, 7687, 7700 and 4222.
What to do.
Only if you are comfortable with a terminal, this shows which program uses the port:
lsof -i :7687What you see. The setup says “Claude Code CLI is required”, or the chat says it has no Claude login.
What to do.
Type this and press Enter:
claudeWindows, 64-bit.
Download Docker Desktop from Docker’s page, run the installer, then start Docker Desktop from the Start menu and wait until it says it is running.
Get Docker DesktopDocker Desktop is open and reports that it is running. It is Docker’s own screen, so it is not reproduced here.
Run the installer and follow the wizard, then start Project Orchestrator from the Start menu. Windows may show “Windows protected your PC” because no signing certificate is set in the build configuration: choose “More info”, then “Run anyway” if you trust the download.
The three problems you are most likely to meet, with what the app shows.
What you see. The app says “Docker Desktop is not running” or “Docker Desktop is required”.
What to do.
What you see. Neo4j or Meilisearch stays “Unhealthy”, or the app says it failed to start services. The services need the local ports 7474, 7687, 7700 and 4222.
What to do.
Only if you are comfortable with a terminal, this shows which program uses the port:
netstat -ano | findstr :7687What you see. The setup says “Claude Code CLI is required”, or the chat says it has no Claude login.
What to do.
Type this and press Enter:
claudeAppImage for Intel/AMD; .deb and .rpm packages for Intel/AMD and ARM64.
Follow Docker’s page for your distribution, then start Docker and wait until it says it is running. The app also recognises Docker installed without the Desktop application.
Get Docker DesktopDocker is running. It is Docker’s own screen, so it is not reproduced here.
For the AppImage: in your file manager, open the file’s properties and allow it to run as a program, then double-click it. For a .deb or .rpm: open it with your software installer, then start Project Orchestrator from your applications menu.
The three problems you are most likely to meet, with what the app shows.
What you see. The app says “Docker Desktop is not running” or “Docker Desktop is required”.
What to do.
What you see. Neo4j or Meilisearch stays “Unhealthy”, or the app says it failed to start services. The services need the local ports 7474, 7687, 7700 and 4222.
What to do.
Only if you are comfortable with a terminal, this shows which program uses the port:
lsof -i :7687What you see. The setup says “Claude Code CLI is required”, or the chat says it has no Claude login.
What to do.
Type this and press Enter:
claude