Know exactly which parts of your codebase are accumulating technical debt — without Jira, without integrations, without config. Just point it at any git repo.
| Feature | Health | Commits | Bug Fixes | Bug % |
|---|---|---|---|---|
| payments | 23 | 112 | 38 | 33.9% |
| auth | 54 | 48 | 12 | 25.0% |
| dashboard | 91 | 67 | 2 | 3.0% |
Engineering managers make refactoring decisions based on gut feeling. faultline gives you the data.
Every sprint, the same features generate support tickets. You can feel it — but you can't prove it to leadership without hard data.
Tickets are inconsistently tagged. Some engineers label everything "bug fix," others don't. Your tooling is only as good as your process.
You want to invest in paying down tech debt, but without a risk map you're guessing which features actually need attention.
One engineer owns the payments feature. Nobody else knows it. faultline surfaces knowledge concentration before it becomes a crisis.
No integrations. No account setup. Just one command.
Run faultline analyze . in any project directory. No setup, no config files, no account.
faultline reads commit messages, changed files, and authors across the past year. Bug fixes are detected automatically by commit message patterns.
Import graph clustering groups files by how they connect. With --llm, Claude maps clusters to business domains — no folder naming conventions required.
A health score per feature (0–100), bug fix ratio, commit trends, and optional flow breakdown — saved as JSON and printed to your terminal.
No YAML files, no account creation, no integrations. Run one command in any git repository and get results immediately. Works on any language — JavaScript, Python, Go, Rust.
faultline analyze .Uses import graph clustering to find connected modules, then Claude (or a local Ollama model) maps them to business features — "payments", "auth", "dashboard" — not technical folders.
--llm --provider anthropic|ollamaEvery feature gets a 0–100 health score based on its bug fix ratio and commit activity. Instantly see which features are critical (red), watch (yellow), and healthy (green).
health = 100 − bug_ratio × 200Go deeper than features. With --flows, faultline breaks each feature into named user journeys — checkout-flow, refund-flow — each with their own risk score.
Average analysis time on
a 1,000-commit repo
Cost to analyze public
and open source repos
Typical Claude API cost
per repo with --llm
Private — your code never
leaves your machine with Ollama
The CLI is free and open source today. The SaaS — with dashboards, GitHub App integration, and team collaboration — is coming soon.