Threadline

Follow data through Python methods, with source and model definitions beside it.

Get started · Documentation · GitHub · Report a bug

Threadline opens your Python repository in a local browser. Choose an endpoint, command, or method to see what data enters it, what it creates or changes, and where values are used or returned. Follow calls across methods while keeping the original code and model declarations close by.

It reads source without importing or running the project being reviewed. That project's dependencies do not need to be installed.

Install and open a repository

Requires Python 3.12 or later. In your Python environment:

python -m pip install --pre threadline-review
threadline review /path/to/your/python-project

Your browser opens the review automatically. If it does not, open the local address printed in the terminal. For virtual-environment setup, see the installation guide.

Choose where to start

Search is available across all three pages. Selecting a method opens its Data flow beside its Data models; nothing is selected automatically. Switch to Steps for the line-by-line explanation or expand Code to read only the selected method. The code pane can also sit beside the analysis at an adjustable width. Related tests expand below the code. Selecting a caller compares both methods' code and flow side by side, with Back returning to your previous view.

Each data link points to source evidence. Unresolved sources and effects remain visible as gaps rather than invented values. The usage guide explains model references, navigation, and the limits of static data flow.

Review a change

Compare a working tree with a Git baseline:

threadline review /path/to/your/python-project --base HEAD

Open Changes to select a changed method or compare its before/after source. Changed files, rename-only records, and edits outside methods remain visible when no callable body changes. Current callers, historical callers, and possible impact are shown separately. These are direct source relationships, not a complete impact analysis.

For source roots, exclusions, snapshot-pinned JSON queries, and evidence retrieval, see the usage guide.

Understand the limits

Threadline uses static analysis. The browser labels each call Calls, Probably calls, Library, or Can't tell according to the source evidence (supported, possible, external, and unknown in JSON output). Common injected, inherited, and constructed receivers can have possible source candidates. Dynamic dispatch can remain unresolved; the view does not establish what happened at runtime. The header shows analysis problems, and Coverage lists parse failures and unmodeled syntax.

This is a beta. Linux and macOS are tested with Python 3.12–3.14; Windows is experimental. Human reviewer sessions remain pending before a stable release.

• Capabilities and limits
• Validation and platform coverage
• Keyboard access and accessibility
• Changelog

Contribute

See the contributor guide for development and testing, and the publishing guide for releases.

MIT License · Security policy