dowsing

Assumes you can't fix everything

Static analysis hands you thousands of equally-weighted warnings. dowsing surfaces only code that is both changed often and complex, and ranks it. Code that is complex but nobody touches is left alone — fixing it pays nothing.

Monorepos and package boundaries are first-class

It reads pnpm / bun / npm / yarn workspaces and Turborepo / Nx / Vite+ directly, and analyses on top of the package graph. That is how it finds packages that always change together while having no dependency between them.

Every formula is published

A Code Health score breaks down to which smell cost how many points, so you can recompute it by hand. Thresholds and weights are labelled as the hypotheses they are, and you can calibrate them per repository.

Verifies that change frequency is real

A git line diff cannot tell whether meaning changed. dowsing compares the AST of each revision to separate real edits from renames, reformatting, and comment-only churn.

Built for coding agents

One --help page is the whole interface, every command speaks JSON, and an optional Agent Skill teaches your agent how to read the numbers. No MCP server to register.

Measures whether it actually helped

dowsing diff re-measures both revisions and reports how much complexity actually moved — for a human commit, an agent's refactor, or a PR in CI. Measured, not predicted.

Thirty seconds

bash

npx dowsing analyze

Run it in a TypeScript monorepo with git history. You get a hotspot ranking, per-package risk, and hidden coupling. There is nothing to configure.

⚠ Top hotspots (complexity=loc):
┌───┬────────────────────────────────┬───────┬────────┬────────┬─────────┐
│ # │ file                           │ score │   freq │     cx │ commits │
├───┼────────────────────────────────┼───────┼────────┼────────┼─────────┤
│ 1 │ services/api/src/runtime.ts    │ 0.999 │ 100pct │ 100pct │     255 │
│ 2 │ services/api/tests/db/…test.ts │ 0.997 │ 100pct │ 100pct │     128 │
└───┴────────────────────────────────┴───────┴────────┴────────┴─────────┘

🔗 Hidden coupling — packages that change together with no dependency (1):
│ @acme/admin-web │ @acme/customer-web │ 33% │ 91 │ hidden ⚠ │

How it differs

CodeScene created this category. dowsing keeps its central insight — not all code is equal — and aims at the places where a general-purpose tool is structurally weak.

	CodeScene	dowsing
Monorepos	Retrofitted as folders	Package graph is first-class
Score provenance	Formulas and thresholds are closed	Fully published, recomputable by hand
Churn quality	Admits renames inflate it	AST comparison removes false churn
Bus factor	Repository or file level	Per package
Fixing	Advice only	Applies, verifies, measures the delta
Languages	39	TS/TSX only — that is where the edge comes from
Price	Paid, per active author	MIT

CodeScene

dowsing

Monorepos

Retrofitted as folders

Package graph is first-class

Score provenance

Formulas and thresholds are closed

Fully published, recomputable by hand

Churn quality

Admits renames inflate it

AST comparison removes false churn

Bus factor

Repository or file level

Per package

Fixing

Advice only

Applies, verifies, measures the delta

Languages

TS/TSX only — that is where the edge comes from

Price

Paid, per active author

MIT

If your codebase spans many languages, CodeScene is the better fit. dowsing trades breadth for precision on TypeScript monorepos.

dowsingFind where fixing pays off