Projects - Cloosphere Guide

A Project is a personal document management unit that bundles a Knowledge Base + chats into one space. When you select a project and chat with the AI, it answers using only that project’s documents. When you share a project, an independent copy is created so each member can use it freely.

Project vs. Knowledge Base

Aspect	Project	Knowledge Base
Purpose	Personal/team document space	Knowledge store for agents
Ownership	Personally owned, copy-based sharing	Workspace-shared
Access	Direct from the sidebar	Connected to an agent for use
Chat	Project-specific chat	Referenced from agent chat
Internal structure	Wraps an auto-generated Knowledge Base	Independent entity

A project stores its documents in an internal Knowledge Base, but that KB does not appear in the workspace Knowledge list. Manage project documents from the project screen only.

Use Cases

Scenario	Description
Work doc management	Group materials per project and ask the AI
Team collaboration	Bundle department documents into a project and share with the team
Research	Organize research materials and analyze with AI
Customer proposals	Manage per-customer requirements and draft with AI
Data analysis	Upload CSV/Excel and let the AI analyze and visualize via Python

Project List

In the Projects section at the bottom of the sidebar, view your projects and projects shared with you.

Project detail screen — Click a project in the list to open it and view projects shared with you

Creating a Project

Create a new project

Hover over the Projects section in the sidebar, then click the ”+” button on the right.

Project creation screen — Select the project type, then enter a name and description.

Enter basic info

Field	Description	Example
Project Type	General / Data Analysis	”General” (default)
Name	Project display name	”2026 Marketing Strategy”
Description	Project purpose (optional)	“Materials for Q1 marketing campaign”

The Data Analysis type uploads CSV/Excel files and analyzes them with Python code. See Data Analysis Projects below.

Done

Click Create Project — you’re auto-redirected to the project detail screen.

File Management

Upload Files

In the project detail screen’s Settings tab, upload files. Uploaded files are auto-vectorized and become AI-searchable.

Project file upload — Drag-and-drop or use the upload button to add files

Drag and Drop
File Picker
Cloud Storage

Drag files directly onto the project area.

Supported file formats:

Category	Formats
Documents	PDF, DOCX, PPTX, TXT, MD
Spreadsheets	XLSX, CSV
Other	When LibreOffice PDF conversion is enabled, additional formats are supported

File List

Uploaded files are visible in the Project Files section under the Settings tab on the project detail screen.

Info	Description
Filename	Uploaded file name
Size	File size
Status	Processing state (uploading / processed / error)

Delete a File

Click the X button on the right of the file. The deleted file is also removed from the vector DB.

File deletion is irreversible. The vector index is also removed — keep originals if needed.

Data Analysis Projects

New feature — Selecting Data Analysis as the project type lets you upload CSV/Excel files and have the AI directly run Python code to analyze data and create charts.

Differences from General Projects

	General Project	Data Analysis Project
File formats	All documents — PDF, DOCX, TXT, etc.	Only CSV, XLSX, XLS, TSV, Parquet
Processing	Text extraction → chunking → vector DB (RAG)	Metadata extraction → mount file in Jupyter
AI response style	Document-search-based answers	Python-execution-based answers
Charts	None	Plotly interactive charts, matplotlib images
Required env	None	Jupyter server connection required

Prerequisites

Data Analysis projects require a Jupyter server to be connected. If a Jupyter server isn’t configured under Admin > Settings > Code Execution, project creation is blocked.

Create and Use

Choose 'Data Analysis' type at creation

On the project creation screen, choose Data Analysis as the type.

Upload data files

Upload CSV, Excel (XLSX/XLS), TSV, or Parquet files. Uploaded files are auto-mounted into the Jupyter environment, and metadata like column info is extracted.

Ask the AI to analyze

Select the project and request analysis in chat. The AI auto-writes and runs Python code, returning results.

Analysis Flow

The AI uses 3 tools sequentially to analyze data.

Step	Tool	Description
1	data_file_info	List uploaded files, columns, data types, etc.
2	get_file_details	Sample detail data for a specific file (head, describe, etc.)
3	code_interpreter	Run Python code in the Jupyter kernel (pandas, plotly, matplotlib, etc.)

Chart Generation

When the AI is asked to visualize data, it generates Plotly interactive charts. Charts can be inspected directly in chat — zoom, hover for info, etc.

Chart Library	Support	Form
Plotly	Default	Interactive (zoom, hover, filter)
matplotlib	Supported	Displayed as PNG image

Example Conversation

Q: Show me the monthly trend in the sales data
A: Analyzing the uploaded sales_2025.csv file.

   [Code Interpreter execution: pandas monthly aggregation + Plotly chart]

   📊 Monthly sales trend:
   [Interactive chart]

   Key findings:
   - March is up 23% MoM
   - Q1 total revenue is ₩12.5M

The Jupyter kernel persists per project. Variables and dataframes created in earlier conversations remain available later. When the Jupyter container restarts, files are auto-remounted.

Chat in a Project

Project-Context Chat

When a project is selected and you start chatting, the AI references only that project’s documents.

Pick a project

Click a project in the sidebar.

Type your question

Type your question in the chat input. The AI auto-searches the project documents.

Review the response

The AI searches the project’s documents and answers with citations.

Project chat — Selecting a project and chatting references only that project's documents

Example:

Q: What's this quarter's marketing budget plan?
A: Based on the project documents, this quarter's marketing budget is...
   [Source: Q1_marketing_plan.pdf, page 12]

Chat Management

Chats inside a project appear in the Chat tab of the project detail screen.

Info	Description
Chat title	Auto-generated conversation title
Modified	Last conversation time

Project chats don’t appear in the main sidebar chat list. They’re separated per project for clean management.

You can share projects with other users. Sharing is copy-based.

Open the Settings tab

On the project detail screen, go to the Settings tab.

Pick share targets

In the Copy to Users section, search for and select users. You can share with multiple at once.

Run the share

Click Copy to Users — a project copy is created for each selected user.

Received project copy — The recipient receives an independent copy, marked '(from [owner])'

Item	Description
Independent copy	A separate project is created for each recipient
File copy	Original project files are copied along with vector re-indexing
Independent edits	After sharing, each user can freely edit their own project
Origin tracking	The copy’s metadata records origin info (owner, project name, copy time)

Sharing is a one-time copy. Edits to the original don’t propagate to copies. Use access-permission sharing on workspace Knowledge Bases when real-time sync is needed.

Project Settings

In the project detail screen’s Settings tab, edit project info.

Item	Description
Name	Rename project
Project Files	Manage and upload files linked to the project
Default model	Pick the default AI model for project chats
Project Instructions	Set a project-specific system prompt for the AI
Copy to Users	Share project copies with selected users
Delete	Permanently delete the project and all linked resources

Deleting a Project

Deleting a project removes all linked resources together.

Project deletion permanently removes:

The project itself
The linked Knowledge Base and vector index
All chat history within the project

Deleted projects can’t be recovered. Recipients’ copies are unaffected.

FAQ

Are there file count limits per project?

Depends on system settings. Typically, no file count limit, but file size limits follow admin settings.

Can I convert an existing Knowledge Base into a project?

Direct conversion isn’t possible. Create a project and re-upload the files.

Do project chats appear in the regular chat list?

No — project chats only appear in the Chat tab of the project detail screen. They’re separated from the main sidebar chat list for clean organization.

Can I see origin info on a shared project?

Yes — the shared project’s metadata records the original owner, project name, and copy time.

What file formats does Data Analysis projects support?

Only CSV, XLSX, XLS, TSV, Parquet. Use a General project for documents like PDF or DOCX.

What environment is needed for Data Analysis projects?

An admin must connect a Jupyter server. The Jupyter server URL must be set under Admin > Settings > Code Execution for project creation. Without Jupyter, choosing the Data Analysis type displays guidance.

Can I reuse variables from earlier conversations in Data Analysis?

Yes — the Jupyter kernel is preserved per project, so variables and dataframes from earlier conversations remain available later.

​Project vs. Knowledge Base

​Use Cases

​Project List

​Creating a Project

​File Management

​Upload Files

​File List

​Delete a File

​Data Analysis Projects

​Differences from General Projects

​Prerequisites

​Create and Use

​Analysis Flow

​Chart Generation

​Example Conversation

​Chat in a Project

​Project-Context Chat

​Chat Management

​Project Sharing

​Sharing Characteristics

​Project Settings

​Deleting a Project

​FAQ

Project vs. Knowledge Base

Use Cases

Project List

Creating a Project

File Management

Upload Files

File List

Delete a File

Data Analysis Projects

Differences from General Projects

Prerequisites

Create and Use

Analysis Flow

Chart Generation

Example Conversation

Chat in a Project

Project-Context Chat

Chat Management

Project Sharing

Sharing Characteristics

Project Settings

Deleting a Project

FAQ