Chronicle records Git commits and Django migrations into a local timeline, turns that history into findings, and lets you ask questions about it with AI — all stored on your machine.
Python 3.10+ · works in any Git repository
Chronicle reads your project, records what changes, and lets you ask questions about the result — in three steps.
Scans Git commits and Django migrations into structured, deduplicated records stored in a local SQLite database.
Analyzers flag interesting signals: newly created files, added or removed database fields, and more.
Ask plain-language questions; Chronicle feeds the most relevant findings and commits to OpenAI or Gemini.
Clone, install, and point Chronicle at any Git repository.
Grab the repository and install the CLI in editable mode.
# clone git clone https://github.com/Sakshyam-Xtha/Chronicle.git cd Chronicle # install (includes the `chronicle` command) pip install -e ".[dev]"
Run init inside any Git project to create .chronicle/ and the local database.
cd /path/to/your/project
chronicle init
# ✓ Chronicle directory created
# ✓ Local database initialized
Record history, then turn it into findings. Both are incremental.
chronicle scan chronicle show chronicle analyze
Configure a provider, export your key, and interpret the timeline.
chronicle config set provider openai chronicle config set model gpt-4o export OPENAI_API_KEY="sk-..." chronicle interpret --question "Why did we remove that field?"
One small, predictable CLI. No background daemons, no web UI required.
chronicle initInitialize Chronicle in a projectchronicle scanCollect observations (git, migrations)chronicle showList observations; --id for detailchronicle analyzeGenerate findings from observationschronicle interpret -q "…"AI answers about the historychronicle configView / set provider & modelchronicle statusShow initialization + Git statechronicle versionPrint the versionDeveloper intelligence that respects your repository and your privacy.
Everything lives in .chronicle/. No cloud, no sync, no telemetry.
Checkpoints remember progress, so scans only process what's new.
Add your own scanners and analyzers; providers for OpenAI and Gemini.
Keyword scoring sends only the most relevant findings and commits to the model.
Observations, findings, and state in one portable chronicle.db.
Fork it, audit it, repurpose it. It's developer tooling, after all.
Record how your code evolves, understand why it happened, and keep the answers forever — right next to the code.