Trovec Viewer for VS Code
A VS Code extension for viewing and querying .trovec vector database files directly in the editor.
Features
- Custom editor — opens
.trovec files with a visual interface instead of raw binary
- MongoDB-style queries — familiar fluent syntax for browsing, filtering, and aggregating data
- Autocomplete — context-aware suggestions for commands, operators, and field paths from your data
- Live reload — automatically refreshes when the
.trovec file changes on disk
Usage
- Open any
.trovec file in VS Code
- The Trovec Viewer opens automatically
- Type a query and press Ctrl+Enter (or click Run)
Query Reference
Commands
db.find() // all entries (paginated)
db.find({ id: "doc-1" }) // find by exact ID
db.find({ id: { $regex: "^doc-" } }) // ID pattern match
db.find({ "context.category": "news" }) // filter by context field
db.find({ "context.score": { $gt: 0.5 } }) // comparison operators
db.count() // total entry count
db.count({ "context.type": "article" }) // filtered count
db.distinct("context.category") // unique values for a field
db.stats() // collection metadata
Chaining
db.find({ "context.type": "article" })
.sort({ "context.date": -1 }) // sort (1 = asc, -1 = desc)
.skip(20) // offset
.limit(10) // max results
Filter Operators
| Operator |
Description |
Example |
| (none) |
Exact match |
{ "context.type": "article" } |
$gt |
Greater than |
{ "context.score": { $gt: 0.5 } } |
$gte |
Greater than or equal |
{ "context.score": { $gte: 0.5 } } |
$lt |
Less than |
{ "context.score": { $lt: 1.0 } } |
$lte |
Less than or equal |
{ "context.score": { $lte: 0.9 } } |
$ne |
Not equal |
{ "context.status": { $ne: "draft" } } |
$in |
In array |
{ id: { $in: ["a", "b"] } } |
$nin |
Not in array |
{ "context.type": { $nin: ["draft"] } } |
$regex |
Pattern match (case-insensitive) |
{ "context.title": { $regex: "^intro" } } |
$exists |
Field exists |
{ "context.tags": { $exists: true } } |
$and |
Logical AND |
{ $and: [{ ... }, { ... }] } |
$or |
Logical OR |
{ $or: [{ ... }, { ... }] } |
Multiple fields in one object are implicitly $and:
db.find({ "context.type": "article", "context.score": { $gte: 0.8 } })
Result Table
- Click any row to expand it and see the full context JSON, embedding values, and vector stats (min, max, mean, L2 norm)
- Results are paginated at 50 rows per page
Development
# Install dependencies
npm install
# Compile
npm run build
# Launch in VS Code
# Press F5 (uses the "Run Trovec Viewer Extension" launch config)
The extension reads .trovec binary files (with optional Brotli compression) produced by @trovec/core. It supports all quantization types (F32, INT8, BIT) and metrics (cosine, euclidean, dot, hamming).
