PYPIv0.1.2Python 3.11+ & GitMIT License

jung

A Version Control Sidecar for AI-Assisted Development with Prompt, Plan, Cost & Diff Attribution

Git tracks WHAT changed. jung tracks WHY and WHO (human or AI agent). A local-first development sidecar that captures conversation logs, multi-step plans, tool executions, token expenditures, and human-vs-agent code attribution behind every code change — permanently linked to git diffs.

View on PyPI RegistryGitHub Repository
ATTRIBUTION ACCURACY
100%
STORAGE ARCHITECTURE
SQLite + SHA-256 Blobs
GIT HOOK SPEED
<15ms
CLI COMMANDS
24+
On This Page
1. Installation2. Quickstart & Usage3. CLI Reference (24+ Commands)FAQ & SecurityRelated Packages

1. Installation

Install jung using your favorite package manager. All dependencies are minimal and zero-leakage.

Install Command
pip install jung

2. Quickstart & Implementation

Get up and running in less than 30 seconds. Copy and paste the code snippets below directly into your codebase.

Code Examples
# 1. Install the CLI sidecar
pip install jung

# 2. Initialize in your repository
cd my-project
jung init

# 3. Start the background watcher
jung watch

# 4. Check status and browse captured AI context
jung status
jung log
jung cost
jung blame src/main.py

# 5. Launch the local web dashboard
jung dashboard --open
File: terminalbash

3. Command-Line Reference

jung provides 24+ commands for tracking AI turns, Git diffs, token consumption, and code attribution.

$ jung init

Initializes a .jung/ store with SQLite index and content-addressed blob store in the repository. Automatically installs a Git post-commit hook.

$ jung watch

Starts the background daemon to discover IDE sessions (Antigravity, Cursor, Claude Code) and correlate file modifications to conversation turns.

$ jung status

Checks daemon health, active branch, last AI-attributed commit, uncommitted changes, and store statistics.

$ jung log

Lists captured sessions, conversation turns, actor attribution, models used, and estimated USD costs.

$ jung show <turn-id>

Displays full turn context: prompt/response content, multi-step agent plans, tool calls, and syntax-highlighted diffs.

$ jung diff [turn-id]

Compares file changes produced in specific conversation turns or across entire AI sessions.

$ jung blame <file>

Like git blame, but shows whether each line was authored by a human or AI agent, plus turn ID and date.

$ jung cost

Summarizes token expenditures and financial costs grouped by session, calendar day, or AI model (Gemini, Claude, GPT).

$ jung search <query>

High-speed full-text search across turns, agent reasoning, and markdown artifacts using SQLite FTS5.

$ jung dashboard --open

Spawns a local-first interactive web dashboard to browse conversation threads, diff lineages, and cost charts.

$ jung commit -m 'msg'

Creates an explicit checkpoint snapshot of AI changes and conversation context.

$ jung branch <name>

Branches development context for isolated AI experimentation without disturbing the main timeline.

Interactive CLI Terminal Simulator

Explore Open-Source CLI Tooling

Target:jung (PyPI)
Sets up local .jung/ SQLite metadata store & Git post-commit hook.
bash — terminal
Local-First Execution
$ jung init
-> Initialized empty jung store in .jung/
-> Database created: .jung/index.db (SQLite WAL mode)
-> Content-addressed blob store created: .jung/objects/
-> Git repository detected: installing post-commit hook...
✔ SUCCESS: Bidirectional Git hook installed. Turns will automatically link to commits.

Frequently Asked Questions