Skip to content

Core concepts

GitIntel has a small conceptual model. Four ideas explain every number it prints.

flowchart LR
    A[Repository source<br/>local path or GitHub URL] --> B[Repository context<br/>owner, branch, remote]
    B --> C[Commit walk from HEAD<br/>+ per-commit diffs]
    C --> D[Derived metrics<br/>summary, contributors,<br/>ownership, hotspots]
    D --> E[Report renderer<br/>table, JSON, Markdown]

  1. A source resolves to a repository context. A local path is opened in place; a GitHub URL is cloned into a cache. Either way GitIntel ends up with an open repository plus metadata. → Repository sources and cache

  2. History is walked exactly once. Commits reachable from HEAD are read in time order and each commit is diffed against its first parent to produce per-file additions and deletions. → Analysis pipeline

  3. Every metric is derived from that same commit list. Contributors, ownership, file activity, and hotspot risk are pure functions over commits and their file changes — the results are cached in memory so nothing is recomputed. → Metrics and scoring

  4. Reports are a presentation layer. The analysis does not know whether it will be rendered as a Rich table, JSON, or Markdown. → Output formats

Read these in order

Page What you learn
Terminology Precise definitions of contributor, owner, churn, hotspot, and more
Analysis pipeline How a command turns a path into a report, and what is lazy or cached
Metrics and scoring Exact formulas and thresholds behind every number
Repository sources and cache Local vs. remote handling, cache location, cleanup rules

If you would rather see the code-level view, the same material is described from the implementation side in Architecture.