git-forensics

分析 Git 历史，发现"逻辑耦合"（总一起改的文件）和"热点"（高频修改的复杂模块）。基于 Adam Tornhill 的《Your Code as a Crime Scene》方法论。

Install with Tessl CLI

npx tessl i github:Lingjie-chen/MT5 --skill git-forensics

What are skills?

1.45x

Quality

56%

Does it follow best practices?

Impact

90%

1.45x

Average score across 3 eval scenarios

Optimize this skill with Tessl

npx tessl skill review --optimize ./.trae/skills/git-forensics/SKILL.md

SKILL.md

Review

Evals

Evaluation results

84%

35%

Finding Hidden Dependencies in an Open Source Service

Temporal coupling analysis

Criteria

Without context

With context

Script output format

30%

Noise files excluded

100%

HIGH_COUPLING label

100%

MEDIUM_COUPLING label

100%

File type classification

20%

70%

Build root annotation

100%

History completeness check

100%

Initial activity survey

100%

Actionable recommendations

100%

Architectural decay identification

50%

40%

Without context: $0.4760 · 6m 5s · 17 turns · 129 in / 8,005 out tokens

With context: $1.2606 · 7m 13s · 37 turns · 3,197 in / 16,537 out tokens

100%

40%

Technical Debt Prioritization for an Engineering Refactoring Sprint

Hotspot detection and debt scoring

Criteria

Without context

With context

Hotspot script output fields

100%

debt_score field present

90%

100%

debt_score = churn × max_ccn

100%

max_ccn used

100%

nloc field present

100%

Sorted by debt_score descending

100%

High churn + high complexity = top priority

86%

100%

High churn + low complexity = deprioritize

91%

100%

Analysis period defined

40%

100%

Concrete file example

100%

Without context: $0.4811 · 3m 49s · 18 turns · 137 in / 6,597 out tokens

With context: $1.2676 · 5m 56s · 39 turns · 8,146 in / 12,871 out tokens

88%

Pre-Decomposition Architectural Risk Assessment

Full forensic report with cross-root coupling

Criteria

Without context

With context

Both analyses present

100%

Cross-subsystem section

100%

Cross-subsystem pairs flagged

100%

Build root annotation

100%

Orphan files listed

100%

One-year orphan threshold

100%

Refactoring priority section

100%

Priority matrix applied

60%

100%

Coupling threshold 0.7

37%

Actionable recommendations

100%

Rename distortion awareness

Without context: $1.5171 · 9m 20s · 37 turns · 281 in / 25,269 out tokens

With context: $1.9167 · 10m 40s · 48 turns · 4,378 in / 24,278 out tokens

Evaluated: 1 day ago
Agent: Claude Code
Model: Claude Sonnet 4.6

Table of Contents

Finding Hidden Dependencies in an Open Source Service Technical Debt Prioritization for an Engineering Refactoring Sprint Pre-Decomposition Architectural Risk Assessment

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.