35 lines (28 loc) · 1.35 KB

Roadmap

Phase 1 — Foundation ✓

GitHub collector with rate-limit fallback and normalized snapshots
Clone manager (HTTPS / SSH)
Domain models (Pydantic)
CLI (portfolio-auditor full-run)

Phase 2 — Analysis ✓

Modular scanners: structure, documentation, testing, CI, cleanliness
Policy-driven scoring engine (YAML weights)
Deterministic reviewer (no LLM dependency)

Phase 3 — Dashboard ✓

Portfolio ranking and shortlist
Streamlit dashboard (overview, repo table, portfolio view, optimizer, redundancy)
ROI-based action optimizer
Historical comparison (snapshot diff)

Phase 4 — In progress

Short term

SQLite exporter — enable ad-hoc SQL queries on audit artifacts
Empirical calibration of effort_units in action_impact_rules.yaml
Improve redundancy detection (beyond TF-IDF: language + structure signals)
Increase test coverage to 80% (currently targeting 72%)

Medium term

Optional LLM review layer (additive, not a replacement for deterministic scoring)
Improved site export (richer JSON payload for portfolio website integration)
Multi-owner support in the dashboard (compare portfolios across accounts)

Long term

Public scoring benchmark against a curated set of reference repositories
Scoring policy versioning (track score changes across policy updates)