Feature: per-language scorecards for mixed-language repositories

[wang93wei]

Summary

Feature request: for multi-language repositories, desloppify should generate per-language score panels/scorecard images in addition to the aggregate codebase score.

Project Link

• Example mixed-language project: https://github.com/wang93wei/SwiftMTP
• Languages in use: Swift + Go

Context

This project contains multiple languages (Swift + Go). We run language-specific scans such as:

desloppify --lang go scan --path Native

But in day-to-day workflow, teams also run whole-repo scans (--path .). In mixed-language repos, a single aggregate score can hide debt in one language while another language remains healthy.

Why this matters (aligned with current tool guidance)

desloppify scan output includes the instruction block for agents:

• "ALWAYS present ALL scores to the user after a scan"
• show overall/objective/strict/verified
• show all mechanical + subjective dimensions

For multi-language repositories, this same principle should apply per language, not only globally.

Current behavior

• Aggregate scorecard is generated (for current scan state).
• Users can force one language via --lang, but there is no first-class multi-language breakdown artifact from one run.

Expected behavior

When scanning a mixed-language repo, provide language-separated reporting artifacts, for example:

1. Per-language score sections in CLI output (Swift/Go/TypeScript/etc), each with:
   • overall/objective/strict/verified
   • mechanical + subjective dimensions
2. Per-language scorecard image files, e.g.:
   • scorecard-go.png
   • scorecard-swift.png
3. Optional aggregate image remains available (scorecard.png).
4. status should expose the same per-language breakdown without requiring separate manual runs per language.

Suggested CLI shape (example)

• desloppify scan --path . --by-language
• desloppify status --by-language
• desloppify scan --path . --badge-path scorecard-{lang}.png --by-language

Acceptance criteria

• In a repo with >=2 detected languages, --by-language outputs distinct score blocks and artifacts per language.
• Each language block follows current score-reporting contract (overall/objective/strict/verified + all dimensions).
• Aggregate + per-language outputs are both available and clearly labeled.
• State interactions are deterministic (no ambiguity between global state and language-scoped state).

Notes

Related reliability friction we also observed while running language-focused review batches is tracked separately in issue #139.

[peteromallet]

Will add this soon @wang93wei

[peteromallet]

Thanks for the feature request @wang93wei. Per-language scorecards for mixed-language repos is a reasonable ask. Currently you can run desloppify --lang go scan --path . and desloppify --lang swift scan --path . separately to get per-language breakdowns, but a first-class --by-language flag would streamline this.

Leaving open as a tracked feature request.