Unlike standard AI commit generators that produce a single sentence, GitSage breaks every analysis into three distinct tactical sections — giving you a complete picture of every change.
Overview
When you run gitsage, the engine produces a structured report with three sections. Each targets a different audience and serves a different purpose.
A concise structural summary of the code modifications. GitSage extracts file paths, added/removed line counts, function signatures, and class-level changes from the raw git diff.
The logical reasoning behind the change. GitSage infers business context from code structure, naming conventions, and diff patterns — providing the justification that reviewers and future developers need.
bash
12345
# GitSage Report — Why it Matters
Prevents stale sessions from remaining active beyond
JWT expiry, eliminating a class of 401 Unauthorized
errors in production. The automatic session invalidation
reduces support tickets from users stuck in bad auth states.
Stealth Active
Ollama Ready
🎯
Reach & Scope
Identification of all modules, classes, and files affected by the change. This helps catch unintended ripple effects and ensures nothing is missed in code review or testing.
Every suggestion includes a Confidence Score (0–100%). If confidence drops below 50%, GitSage will prompt you for manual input rather than committing an uncertain message.
[████████████████████░░] 91% — High confidence. Committing...
[██████████░░░░░░░░░░░░] 47% — Low confidence. Please review before committing.
[████░░░░░░░░░░░░░░░░░░] 18% — Very uncertain. Manual commit message required.