Problem
The most compelling section in every report was the cross-repo comparison table (React vs Effect-TS vs uv). Built entirely by hand — copying numbers between reports and formatting a comparison table.
Proposal
```bash
vajra compare /data/react/ /data/effect/ /data/uv/ --labels 'React,Effect-TS,uv' --format markdown
```
Output:
| Metric |
React |
Effect-TS |
uv |
| Commit entropy |
3.71 |
3.45 |
3.07 |
| Author cardinality |
66 |
88 |
44 |
| Fix ratio |
22.8% |
17.1% |
19.5% |
| Community merge rate |
10.0% |
45.9% |
30% |
| One-commit rate |
59.1% |
59.1% |
81% |
| Bus factor (80%) |
4-10 |
1-2 |
3-8 |
Also computes pairwise drift between all repos and ranks them by similarity.
Implementation
- Takes N directories, each containing pre-computed vajra outputs
- Or N JSON files from `vajra ingest` / `vajra score`
- Produces comparison table + pairwise drift matrix
Evidence
Built this table manually for 3 reports. It was the most shared section.
Complexity
Medium. Orchestration over existing analysis, plus tabular output formatting.
Problem
The most compelling section in every report was the cross-repo comparison table (React vs Effect-TS vs uv). Built entirely by hand — copying numbers between reports and formatting a comparison table.
Proposal
```bash
vajra compare /data/react/ /data/effect/ /data/uv/ --labels 'React,Effect-TS,uv' --format markdown
```
Output:
Also computes pairwise drift between all repos and ranks them by similarity.
Implementation
Evidence
Built this table manually for 3 reports. It was the most shared section.
Complexity
Medium. Orchestration over existing analysis, plus tabular output formatting.