0+
Result limit ceiling
Desktop intelligence for maker libraries
ForgeIndex 1.0
Build a clean, searchable, isolated library for every drive. Scan faster, review smarter, and enrich files with offline vision plus an optional manual AI lane that stays separate until you explicitly review and promote it.
- Windows 10/11 + .NET 8
- Branded startup splash + Windows 11 icon
- Library-isolated SQLite workspaces
- Offline ONNX/PT vision pipeline
- Optional per-library AI add-ons
- Health, integrity, and audit dashboards
0
Scan passes
0
Core domains
0.0
Current release track
Capability map
Advanced features that ship with ForgeIndex
Use filters to inspect the stack by operational lane.
Multi-library segregation
Each library keeps its own DB and metadata inside .forgeindex. No cross-talk, no mixed rows.
Drive hotplug discovery
Attach drives at runtime, detect existing manifests, and continue from known libraries instantly.
Offline ONNX/PT inference
Select .onnx or .pt. PT models auto-export and cache to ONNX before inference.
Two-phase pending review
Stage detections, bulk-edit labels, then commit or discard transactionally with replace safety guardrails.
Optional AI add-on lane
Keep AI fully opt-in with global and per-library enablement, manifest-backed pack discovery, and honest readiness state.
Manual AI review and promotion
Run selected-image AI analysis manually, review stored results, then explicitly promote accepted labels into normal tags or map one category.
Health + quality dashboards
Find roots with missing cover, missing models/images, stale rows, broken links, and integrity drift.
Alias + signature intelligence
Repair generic root names, detect duplicate structures, and enrich mesh metadata for safer triage.
Installed tool auto-discovery
Discover installed slicer/CAD/EDA/vector/image tools from common install roots and Windows registry metadata.
Extension-aware launch routing
Assign extension profiles per tool so Open in <Tool> picks the best configured app by file type.
Performance diagnostics
Track query latency, decode queues, stall warnings, enrichment throughput, and runtime status in-app.
Operational audit trail
Every critical action is timestamped and queryable for transparency, rollback planning, and reporting.
Release channel
Download ForgeIndex v1.0
Primary build
Use this package for first-time installs and clean upgrades.
Download ForgeIndex v1.0What users need
- Windows 10 or Windows 11
- .NET 8 runtime/SDK
- Optional NVIDIA CUDA stack for GPU vision acceleration
- Optional local AI packs if you plan to use the AI add-on lane
- At least one maker-library root folder or drive
Quick start
- Install and launch ForgeIndex.
- Create or attach a library in the Libraries menu.
- Run initial scan.
- Use Preferences -> External Tools -> Discover Installed Tools.
- Optionally configure Vision Control Center, AI Add-ons, and dictionary layers for library-scoped enrichment workflows.
Onboarding
Tutorial walkthrough
Use the video to learn setup, scan strategy, offline vision review, optional AI review/promotion, and quality maintenance loops.
Walkthrough coverage
- Library creation and storage policy
- Scan monitor and extension telemetry
- External tool discovery and extension routing
- Vision profile selection and pending review
- Selected-image AI review and explicit promotion
- Health dashboard cleanup workflow
- Reports and backup strategy
Community repository
Download add-in dictionary packs
Users can layer dictionaries by library without touching the shipped base dictionary.
Place downloaded JSON packs under:
<library-root>\.forgeindex\dictionaries\
then reload dictionary layers in ForgeIndex.
FAQ
Common launch questions
Can I keep separate libraries on separate drives?
Yes. ForgeIndex is designed for library-local metadata with strict isolation guardrails.
Can users run vision offline?
Yes. Vision enrichment is fully offline and supports ONNX plus PT-to-ONNX export workflow.
Is the newer AI add-on lane required?
No. ForgeIndex core works normally with zero AI packs installed. The AI lane is global/per-library opt-in and remains additive only.
Does AI auto-write tags or categories?
No. Stored AI annotations stay separate until you explicitly review and promote a label into a normal tag or map one category.
Do dictionary packs overwrite user tags?
No. Manual tags remain untouched. Dictionary inference applies controlled auto-tags with confidence metadata.