Historical deployment analysis & Code Intelligence — walk git history day by day, deploy per commit, test all endpoints, capture screenshots, and analyze code evolution to find duplicates, rank quality, and generate refactor plans.
SUMD stats (v0.1.23): 3722 functions · 161 classes · 167 files · CC̄ = 3.9
- 🤖 LLM usage: $96.3548 (63 commits)
- 👤 Human dev: ~$2159 (21.6h @ $100/h, 30min dedup)
Generated on 2026-07-06 using openrouter/qwen/qwen3-coder-next
| Document | Description |
|---|---|
| Getting Started | Installation, quickstart, configuration |
| Walk Command | Core walk workflow — deploy modes, output, rebuild.yaml |
| Analyze | Duplicates, service graph, truth, vector search, multi-repo |
| Refactor | AI-powered refactor plans + execution |
| Auto-PR | Automated GitHub/GitLab PR creation |
| Plugins | Custom scanners & reporters via entry points |
| CLI Reference | All commands and options |
| Config Reference | Full rebuild.yaml field reference |
| Architecture | 5-layer design, services, data flow |
| Case Study: c2004 | Real-world analysis of a large ecosystem |
| c2004 Testing Log | Detailed test log with commands & results |
| Usage Guide | Step-by-step guide (all features) |
| Changelog | Release history (current: v0.1.23) |
| Roadmap | Completed phases 10–16, future plans |
| Analysis (P1–P4) | Refactoring hotspots, infra gaps, missing features, test debt + sprint plan |
# Install
pip install rebuild
# Initialise config
rebuild init /path/to/my-service
# Walk last 14 days (dry-run, no deploy)
rebuild walk /path/to/my-service --deploy none --days 14 --dry-run
# Serve results
rebuild serve --results-dir /path/to/my-service/.rebuildSee Installation and Quick Start for details.
| # | Layer | Description |
|---|---|---|
| 1 | Intelligence | Structural & semantic duplicates (Python/JS/TS), service graphs, vector search, quality ranking |
| 2 | Decision Engine | AI-powered refactoring plans + automated execution |
| 3 | Walk | Git history iteration (Incremental, Replay, Accelerator modes) |
| 4 | Deploy | Docker Compose / Replay / Accelerator hot-reload per commit |
| 5 | Scan & Test | OpenAPI, FastAPI routes, Traefik discovery; auth, param substitution, body injection |
| 6 | Visualization | D3.js graphs, health dashboards, SSE streaming, evolution playback |
| 7 | Automation | Auto PR, DSL scripting, NLP commands, Plugins, Notifications |
Applying rebuild to the massive c2004 project (88 subdirectories, thousands of files).
See full details: Case Study · Testing Log
rebuild analyze duplicates /path/to/c2004
Result: Found 1,597 duplicate groups across Python and JS/TS files.
rebuild analyze services /path/to/c2004/backend --export
Result: Interactive D3.js map highlighting circular dependencies.
rebuild refactor plan /path/to/c2004/backend --ai
rebuild refactor pr /path/to/c2004/backendResult: 122 high-impact suggestions with automated PR descriptions.
rebuild dashboard --repo /path/to/c2004
Result: Correlation between technical debt and API pass rates.
Explore ready-to-run scenarios in examples/:
- 01-dry-run-walk: Standard walk + intelligence.
- 02-docker-compose-project: Full pipeline with Docker isolation.
- 03-restore-endpoint: Discovering "truth" and extracting endpoints.
All Phases 10–16 completed. See TODO.md for full roadmap.
| Milestone | Status |
|---|---|
| c2004 Testing & Intelligence (Phase 10) | ✅ Done |
| Semantic Embeddings & Vector Search (Phase 11) | ✅ Done |
| Multi-Repo & Real-time (Phase 12) | ✅ Done |
| UI/UX (Phase 13) | ✅ Done |
| Production Readiness (Phase 14) | ✅ Done |
| c2004 Integration (Phase 15) | ✅ Done |
| Production Release (Phase 16) | ✅ Done — PyPI, CI/CD, Docker, MkDocs, Plugins, Notifications |
Current: v0.1.23 · 634 tests · 72% coverage
Licensed under Apache-2.0.