---
name: "vector-index-health-check"
description: "Diagnose a vector index using supplied IDs, source metadata and similarity samples. Use when the user needs: Get separate duplicate, orphan and freshness findings with reproducible queries."
metadata: {"version": "2026-10-17.1", "category": "search-retrieval", "publisher": "AIDiveForge"}
---

# Vector Index Health Check

Get separate duplicate, orphan and freshness findings with reproducible queries.

## Requirements

Provide the index schema, source manifest, timestamps and read access or an exported sample.

## Workflow

1. Record embedding model/version, chunk identifiers, source IDs and the expected update policy.
2. Compare index source IDs with the authoritative source manifest to identify definite orphans.
3. Compare source revision/update markers with index markers to identify stale chunks.
4. Inspect high-similarity candidates before labeling them duplicates; choose thresholds from the actual embedding/model context.
5. Return read-only diagnostic steps and a repair proposal; do not delete or re-embed records without authorization.

## Worked example

Read `examples/input.txt` and `examples/expected.txt` before adapting the workflow.

## Acceptance checks

Classify a as stale and b as orphan; do not infer duplicates without the necessary data.

## Limits

A sample may miss issues. Similarity alone does not prove semantic duplication or justify deletion.

Worked example included. Model-specific performance and vendor compatibility have not been benchmarked.