# Agent Evolution Archive

This directory contains temporary scripts, reports, and artifacts from the April 2026 model research and data migration work.

## Structure

- `scripts/` — One-off Python/Node scripts used for data migration, diagnostics, and fixes
- `reports/` — Intermediate analysis outputs (JSON, Markdown, text)
- `data/` — Test data files and old dashboard builds

## Notable Scripts

| Script | Purpose |
|--------|---------|
| `add_current_model_id.py` | Adds `current_model_id` string field to agent scores (replaces fragile index lookup) |
| `add_fallback_models.py` | Injects `fallback_models` list into `model-benchmarks.json` agents |
| `add_groq_fallbacks.py` | Adds Groq provider fallback configuration with rate limits |
| `analyze_if_scores.py` | Analyzes IF (Inverse FLOPs) scores across model matrix |
| `check_fixes.py` | Validates fixes applied to `model-benchmarks.json` |
| `fix_model_benchmarks.py` | Corrects field name mismatches (`before`/`after` vs `score_before`/`score_after`) |
| `update_failover_strategies.py` | Adds `failover_strategy` blocks to capability-index.yaml |
| `validate_fixes.py` | Full validation suite for `model-benchmarks.json` structure |
| `verify_fallback.py` | Checks fallback model existence in benchmark data |
| `verify_fixes_simple.py` | Quick structure verification after fixes |
| `_validate_benchmarks.py` | JSON schema validation runner |
| `rebuild-template.js` | ES module version of `rebuild-template.cjs` (superseded) |
| `sync-benchmarks-from-yaml.js` | ES module version of sync script (superseded) |

## Reports

| Report | Description |
|--------|-------------|
| `analysis_results.json` | IF score analysis output |
| `optimization_summary.md` | Model assignment optimization recommendations |
| `recommendations_summary.md` | Condensed recommendation matrix |
| `research_report.md` | Full research findings before dashboard integration |
| `quick_findings.txt` | Bullet-point summary for quick reference |
| `patch.json` | Proposed data patches before application |

## Data Files

| File | Description |
|------|-------------|
| `model-research-test.json` | Test fixture for dashboard rendering |
| `research-dashboard-2026_04_27.html` | Previous dashboard build (superseded by 2026_04_29) |

## When to Reuse

These scripts may be useful for:
- Future data migrations when `model-benchmarks.json` schema changes
- Re-adding fallback configurations after resets
- Benchmarking new model providers
- Diagnostic checks when dashboard fails to render

## Migration Notes

All fixes applied by these scripts have been merged into the production data:
- `model-benchmarks.json` contains `current_model_id` on every agent
- `capability-index.yaml` contains `fallback_models` and `failover_strategy`
- Field names standardized to `before`/`after` (not `score_before`/`score_after`)