Project posture
Local-first storage analysis and recommendation system with strict read-only constraints.
Storage Strategist | Hosted project
Read-only storage intelligence with explicit safety boundaries.
Storage Strategist is a local-first storage analysis toolchain built around a Rust core, CLI, service facade, and desktop review path. The useful signal is not only disk scanning; it is the discipline around read-only behavior, explainable recommendations, parity checks, diagnostics, and a UI that keeps review separate from destructive file operations.
Project snapshot
Use the snapshot to understand scope, build posture, and current focus quickly.
This section stays compact on purpose so both product pages and deeper technical builds can start with the same fast read.
Core stack
Rust workspace with core, CLI, service facade, synthetic fixtures, and Tauri + React desktop scaffold.
Safety model
No delete, move, rename, or modify operations; recommendations stay advisory and policy-gated.
Quality signal
Workspace tests, clippy, compliance checks, desktop smoke tests, KPI gates, and benchmark workflows.
Build story
Follow the project through a small number of deliberate chapters.
Each chapter can lean more product-first or more systems-first, but the route structure stays stable across both.
The project starts from a hard boundary: analysis only, never file mutation.
Storage tools become risky when recommendations and destructive actions live too close together. Storage Strategist keeps the engine local and read-only, records permission and metadata problems as warnings, and excludes unsafe target classes from optimization decisions.
The scanner and recommendation engine do not perform delete, move, rename, or modify operations.
Cloud, network, virtual, and OS-related mounts are analyzed but excluded as local optimization targets.
Permission and metadata failures are downgraded to warnings so scans can continue best-effort.
Policy traces make blocked recommendations visible instead of hiding important safety decisions.
The Rust core treats reports, recommendations, and diagnostics as stable contracts.
The implementation is split into a scanning and analysis core, a CLI, and a service layer for UI consumers. That shape makes the project stronger than a one-off disk usage script because reports, policy decisions, and diagnostics can be reused across interfaces.
- The core owns scanning backends, disk enrichment, category inference, duplicate detection, recommendation policy, and report schema.
- Report versioning is additive by default so CLI, service, and desktop consumers can evolve against a stable contract.
- Diagnostics bundles combine report data, doctor output, and environment metadata for support-style review.
- Scenario planning remains a read-only what-if layer built from policy-safe recommendations.
The desktop direction is a review workbench, not an automation panel.
The Tauri and React scaffold exposes guided path selection, scan progress, result tabs, rule traces, scenarios, and diagnostics without turning the product into a file-operation tool. The service facade gives the UI a clean boundary for sessions, events, reports, recommendations, and export.
- The UI flow starts with guided path selection before scanning begins.
- Scan sessions emit progress events, counters, warnings, and cancellable state through the service layer.
- Results are organized into disks, usage, categories, duplicates, scenarios, recommendations, and rule trace views.
- Diagnostics export supports review and troubleshooting without granting mutation controls.
The credibility comes from gates around parity, performance, and recommendation behavior.
Storage Strategist is useful portfolio proof because it shows engineering around uncertainty: backend parity, benchmark regression checks, recommendation KPI thresholds, compliance checks, and desktop smoke tests all shape the work.
- Native and parallel-disk-usage-backed scanning paths are compared through explicit parity metadata.
- Evaluation fixtures track precision, contradiction rate, and unsafe recommendation counts.
- Benchmark workflows watch for performance regressions before backend defaults change.
- Compliance and provenance checks keep dependency and license choices visible.
Proof points
Public proof in the repo shape
The strongest signal is the combination of systems design, safety policy, typed report contracts, and UI direction. The project shows how to build a useful desktop-adjacent tool without crossing into risky automation.
Read-only guarantee
README and architecture notes explicitly prohibit delete, move, rename, and modify operations.
Architecture depth
Separate core, CLI, service, and desktop layers keep analysis, commands, session state, and review UI decoupled.
Recommendation discipline
Policy rules, blocked traces, scenario planning, and KPI gates make advice explainable and testable.
Delivery posture
CI covers formatting, clippy, workspace tests, compliance checks, benchmark regression, and desktop smoke tests.
Next routes
Use the live links, repo, and related site routes next.
Hosted project pages should be able to point both outside the site and back into the broader proof system without requiring a different route pattern per project.
Continue in the site