Skip to content

Reports and baselines

What it is

A CodeClone report is the structural analysis of a repository produced by a single run: clones, design risks (cohesion, complexity, coupling), dead code, coverage, and the health score, in a deterministic form — the same repository state at the same commit always produces the same report.

A baseline is a stored snapshot of a report from a known-good state, typically a stable commit on the main branch. Once set, CodeClone compares new runs against it to mark findings as "novel" or "known" and to track health-score improvement or regression.

Why it exists

A report alone only describes the present; it cannot say whether things got better or worse without something to compare against. Baselines exist to give every subsequent run a fixed point of reference, so that "500 findings" can mean either "500 pre-existing findings, none new" or "500 findings, 12 of them new regressions" — a distinction that matters enormously for CI gating and is invisible without a baseline.

How it fits together

Format Purpose
JSON Canonical, machine-readable; source for all other formats
HTML Human-browsable report with drill-down
Markdown Human-readable summary, e.g. for a PR description
SARIF Static-analysis interchange format for CI code-scanning integrations
Text Plain-text summary for terminals and logs 
graph LR
    A["Analysis run"] --> B["Report (JSON canonical)"]
    B --> C["HTML / Markdown / SARIF / text"]
    B --> D["Set as baseline"]
    D --> E["Future runs compare against it"]

A report is tied to its repository root; comparing reports generated from different roots is meaningless even if the content looks similar. Moving the baseline too often (on every commit) removes the ability to spot regressions — a baseline should move only after a deliberate review, not automatically.