# Optimize Knowledge Graph Entity Presence

Inconsistent entity facts weaken search and AI recognition. This skill audits evidence, resolves identity conflicts, and records authorized canonical updates with provenance.

## Install

```bash
npx skillstore add aaron-he-zhu/entity-optimizer
```

## Metadata

- Status: approved
- Slug: aaron-he-zhu-entity-optimizer
- Version: 17.0.0
- Author version: 17.0.0
- Skillstore revision: r1
- Version status: valid
- Tree hash: c30e634d87e12c57125f0c249d7d013ed0dd7b95dcc6e607eb9d92594ae4aa9d
- Author: aaron-he-zhu
- GitHub username: aaron-he-zhu
- License: Apache-2.0
- Repository: https://github.com/aaron-he-zhu/aaron-marketing-skills/tree/main/protocol/entity-optimizer
- Ref: d71c7417a35d5c2624161bd2fe8de8a41a362128
- Supported tools: Claude, Codex, Claude Code
- Audit status: complete
- Agent install advisory: allowed
- Manual install advisory: allowed
- Artifact signature: available
- Audit attestation: unavailable
- Human verification: not\_verified
- Risk factors: filesystem, external\_commands, network
- Quality score: 78
- Quality tier: bronze
- Public page: https://skillstore.pages.dev/skills/aaron-he-zhu-entity-optimizer
- Manifest: https://skillstore.pages.dev/api/skills/aaron-he-zhu-entity-optimizer/manifest

## Capabilities

- Audits structured data, knowledge bases, identity consistency, corroboration, and AI recognition signals.
- Reconciles duplicate or ambiguous entities using verified cross-links and user confirmation.
- Tracks canonical entity types, aliases, domains, QIDs, sameAs links, and disambiguation evidence.
- Distinguishes unknown observations from failed or partial checks.
- Reviews proposals and records authorized accept, reject, or upsert events through the registry runtime.
- Produces prioritized actions with sources, dates, confidence limits, revisions, and unresolved conflicts.

## Use Cases

- Audit an organization: Assess machine-facing identity signals and prioritize evidence-backed corrections across owned and authoritative sources.
- Reconcile registry proposals: Review proposed facts, resolve duplicate identifiers, and preserve conflicting evidence before authorized acceptance.
- Diagnose AI confusion: Compare entity descriptions and associations to identify why AI systems confuse similar organizations or products.

## Prompt Templates

### Run a basic entity audit

```
Audit entity recognition for [entity name]. Use [official domain] and these verified profiles: [links]. Mark unavailable checks as unknown.
```

### Check identity consistency

```
Compare the canonical name, type, domain, schema @id, QID, and sameAs links for [entity]. Report conflicts and missing evidence.
```

### Resolve ambiguous entities

```
Disambiguate [entity A] from [entity B]. Require verified cross-links, preserve conflicting evidence, and propose changes without writing them.
```

### Adjudicate registry proposals

```
Review pending proposals for [aggregate ID] in offset order. Validate provenance, revision expectations, authority, and downstream dependencies before recommending decisions.
```

## Limitations

- Requires the verified registry runtime to persist canonical changes.
- Does not create page copy, implement schema markup, or define brand positioning.
- Cannot guarantee a Knowledge Panel, Wikipedia eligibility, search ranking, or AI recognition.
- Optional recognition checks may be unavailable and must remain unknown.

## Best Practices

- Use primary sources, record observation dates, and preserve provenance for every factual claim.
- Require verified cross-links or explicit confirmation before merging similar entities.
- Ask before persistent writes and keep capability values out of prompts, files, and logs.

## Anti Patterns

- Do not treat missing optional checks as failures or partial results.
- Do not manufacture notability, citations, profiles, or evidence.
- Do not edit canonical event streams or generated projections by hand.

## Security Audit

- Audited at: 2026-07-12T12:20:07.333\+00:00
- Summary: All 36 static findings are false positives caused by Markdown formatting, documentation links, metadata URLs, or bounded commands with fixed executables and quoted arguments. No prompt injection, credential exposure, unsafe dynamic execution, or malicious data-handling intent was found.

## Stats

- Views: 0
- Downloads: 7
- Favorites: 0
- Popularity score: 0