Estimate change impact before choosing which tests to run.
The demo is captured from the real CLI using synthetic data. Transcript · regenerate with python3 scripts/demo.py.
A small edit can affect files you never opened. diffblast combines local import relationships and Git co-change history to suggest where to look next.
| Start with | Good for | Where diffblast helps |
|---|---|---|
git diff |
See changed lines | diffblast also estimates related files and tests. |
| Full test suite | Validate behavior | diffblast suggests review targets; it does not execute tests. |
git clone https://github.com/LucasXlab/diffblast.git && cd diffblast
python3 -m pip install .
python3 scripts/demo.pyInstall from this checkout to use the reviewed source version. The installed tool has no third-party runtime dependencies; cloning and installation can require network access.
Extract imports and includes from tracked files, then follow reverse dependency edges. Resolution is heuristic, not a compiler.
Include files that repeatedly changed together. Disable this signal with --no-cochange.
List impacted test files and source/test filename matches. Recommendations do not prove that other tests are safe to skip.
Print a terminal report, emit JSON, or write Markdown for review.
diffblast --path /path/to/project --base main
diffblast --json
diffblast --md --out report.md
diffblast --no-cochangeflowchart LR
A[Git diff and tracked files] --> B[Import graph]
A --> C[Co-change history]
B --> D[Reverse reachability]
C --> D
D --> E[Test suggestions and reports]
Reads the selected local repository and Git history. It makes no application network calls. Exported reports contain repository file paths; review them before sharing. See PRIVACY.md for data boundaries.
python3 -m pip install -e ".[test]"
python3 -m pytest -qThe demo uses a temporary fixture and does not inspect your personal repositories or agent homes.
- Improve import resolution across ecosystems.
- Add documented CI recipes and structured output formats.
Read CONTRIBUTING.md. Report reproducible problems in Issues. Follow SECURITY.md for sensitive reports.
MIT.

