Your AI conversation diary — organized, searchable, and local.
Turn scattered chats into clean, structured, browsable digital diaries.
For an optional visual walkthrough, see the FURChats portfolio site.
FUR: AI Conversation Archiving and Retrieval System
FUR is a command-line system that transforms fragmented AI chats into structured, navigable, and fully searchable local archives.
Designed for researchers, developers, and writers who think with AI and need durable memory, clarity, and fast retrieval.
AI chats vanish.
They get buried, lost, unsearchable, and unrecoverable.
FUR turns those conversations into structured knowledge you can return to, reuse, and build on.
FUR stores every AI conversation in a clean local directory.
Each project (a diary) contains multiple conversations, and each conversation contains timestamped messages, metadata, trees, timelines, and optional Markdown attachments.
Core capabilities:
- Local, transparent JSON storage
- Full-project search (
fur search) - Conversation tagging
- Timelines and message trees
- Jot mode and chat import
- Markdown attachments and export
.frsconversation scripting- Fast, portable, offline
cargo install fur-clicargo install --path . --force.fur/
├── index.json # master index
├── avatars.json # persona aliases
├── threads/ # per-conversation metadata
└── messages/ # per-message storage
All files are plain JSON or Markdown.
| Command | Description |
|---|---|
fur new <name> |
Create a conversation |
fur jot "<text>" |
Add a short message |
fur jot "<text>" --file notes.md |
Add a message with a Markdown attachment |
fur chat [avatar] |
Add long-form content |
fur msg |
Edit or delete a message |
| Command | Description |
|---|---|
fur convo |
List conversations |
fur convo <id> |
Switch active conversation |
fur timeline |
Chronological timeline |
fur tree |
Message tree |
fur jump <message-id> |
Jump to a message |
| Command | Description |
|---|---|
fur convo --tag research |
Add a tag |
fur convo --tag "deep learning" |
Add spaced tag (normalized) |
fur convo --clear-tags |
Remove all tags |
fur convo --delete <id> |
Permanently delete a conversation |
fur clone [-i <id>] [--title <name>] |
Deep-clone a conversation (full copy w/ Markdown) |
fur search <query> |
Full-project search |
fur search "deep learning, optimization" |
Multi-query search |
| Command | Description |
|---|---|
fur printed |
Export current thread to Markdown |
fur save |
Export as .frs script |
fur gsearch |
Scan all FUR journals on disk |
# Create a project
mkdir research && cd research
fur new "GPT-5 Experiments"
# Add short notes
fur jot "Symbolic regression tests using KAN"
# Add longer content
fur chat gpt5
# Attach markdown notes
fur jot "Derivations" --file derivations.md
# Explore
fur convo
fur timeline
fur tree
# Search the entire archive
fur search "memory architecture"
fur search "deep, learning"
# Export
fur printed
fur save session.frsFUR’s search engine inspects:
- Message text
- Attached Markdown files
- All conversations across the diary
- Flexible multi-query syntax
- Contextual snippet extraction
Examples:
fur search "universal approximator"
fur search "symbolic regression, metadata"
fur search "deep, learning"Tags are stored at the conversation level:
fur convo --tag research
fur convo --tag "neural forecasting"
fur convo --tag "macro-modeling, metadata-tools"
fur convo --clear-tagsNormalization:
- lowercase
- trimmed
- spaces → hyphens
Example:
Input: deep learning
Stored as: deep-learning
FUR is not a chat client. It is a durable memory system for people who think and work with AI as part of their intellectual workflow.
Principles:
- Local ownership
- Transparent formats
- Reliability
- Portability
- Speed
- Long-term retrieval
Future enhancements include:
- Advanced editors for message modification
- Enhanced search output formats
- New export templates
- Optional encrypted diaries
- Cross-platform installers
MIT License