# SuilaAI Creator TrustRank Skill

> Trust scoring for content creators. Returns a FICO-style score (300-850)
> with full signal breakdown across 26 behavioral signals and 4 pillars.

## Tool: get_creator_trustrank

Score any creator on YouTube, TikTok, or Twitter/X.

### Endpoint

```
POST https://api.suilaai.com/api/v1/demo/analyze
Content-Type: application/json
```

### Input

```json
{
  "handle": "mkbhd",
  "platform": "youtube"
}
```

| Field      | Type   | Required | Description                                      |
|------------|--------|----------|--------------------------------------------------|
| `handle`   | string | yes      | Creator username (with or without @)              |
| `platform` | string | yes      | One of: `youtube`, `tiktok`, `twitter`            |

### Output

```json
{
  "handle": "mkbhd",
  "display_name": "Marques Brownlee",
  "platform": "youtube",
  "trustrank": 736,
  "trust_band": "trusted",
  "trust_band_display": "Good",
  "tier": "MACRO",
  "followers": 19500000,
  "pillars": {
    "P1_provenance": 0.82,
    "P2_temporal": 0.79,
    "P3_semantic": 0.88,
    "P4_lineage": 0.71
  },
  "signals": {
    "S1": 0.95, "S2": 0.88, "S3": 0.72, "S4": 0.68, "S5": 0.81, "S6": 0.57, "S7": 0.45,
    "S8": 0.81, "S9": 0.79, "S10": 0.76, "S11": 0.83, "S12": 0.65, "S13": 0.88,
    "S14": 0.91, "S15": 0.77, "S16": 0.87, "S17": 0.85, "S18": 0.72, "S19": 0.91, "S20": 0.68,
    "S21": 0.68, "S22": 0.73, "S23": 0.57, "S24": 0.52, "S25": 0.74, "S26": 0.65
  },
  "signals_computed": 26,
  "data_source": "api",
  "analyzed_at": "2026-02-19T10:30:00Z",
  "isLive": true
}
```

### Signal Reference (26 Signals)

**Provenance (S1-S7)**

| Signal | Name                        | What it measures                                     |
|--------|-----------------------------|------------------------------------------------------|
| S1     | Identity Verification       | Platform verification status and identity proof      |
| S2     | Audience Authenticity        | Real vs fake follower ratio (HypeAuditor)            |
| S3     | Audience Geography           | Audience location vs claimed location alignment      |
| S4     | Audience Overlap Entropy     | Diversity of audience across platforms               |
| S5     | Ad Disclosure Compliance     | FTC disclosure rate on sponsored content             |
| S6     | External Reputation (KG)     | Knowledge graph edges and external citations         |
| S7     | Content Provenance (C2PA)    | C2PA metadata verification on media assets           |

**Temporal (S8-S13)**

| Signal | Name                        | What it measures                                     |
|--------|-----------------------------|------------------------------------------------------|
| S8     | Follower Growth Stability    | Organic vs suspicious growth patterns                |
| S9     | Posting Frequency            | Content cadence and schedule consistency              |
| S10    | Engagement Consistency       | Like/comment ratio stability over time               |
| S11    | Engagement Stability         | Weekly engagement variance (low = authentic)         |
| S12    | Content Format Entropy       | Feature utilization mix (Reels, Stories, Posts)       |
| S13    | Freshness & Recency          | Days since last post, exponential decay              |

**Semantic (S14-S20)**

| Signal | Name                         | What it measures                                    |
|--------|------------------------------|-----------------------------------------------------|
| S14    | Content Quality (Text)       | Readability, informativeness, NLP analysis           |
| S15    | Content Quality (Visual)     | Visual + audio + editing quality via CV models       |
| S16    | Hashtag Hygiene              | Spam/banned tag ratio vs relevant tags               |
| S17    | Brand Safety                 | Toxicity, hate speech, NSFW content scoring          |
| S18    | Cross-Platform Consistency   | Profile embedding similarity across platforms        |
| S19    | Language Localization Fit    | Language match between creator and audience           |
| S20    | Topic Authority              | Niche expertise depth via GPT-4 + SparkToro          |

**Lineage (S21-S26)**

| Signal | Name                        | What it measures                                     |
|--------|-----------------------------|------------------------------------------------------|
| S21    | Backlink Authority           | Domain authority of inbound links (Moz)              |
| S22    | Collaboration Graph          | Cross-creator reference and collaboration network    |
| S23    | Content Attribution          | Proper attribution links in content                  |
| S24    | SEO Schema Coverage          | JSON-LD structured data completeness                 |
| S25    | Data Completeness            | Connected accounts and profile completeness          |
| S26    | AI Visibility Index          | LLM citation rate across ChatGPT, Claude, Perplexity|

### Pillar Weights (Creator vertical)

| Pillar       | Weight | What it covers                              |
|--------------|--------|---------------------------------------------|
| Provenance   | 16.15% | Identity, authenticity, platform verification |
| Temporal     | 47.41% | Engagement timing, posting patterns, growth  |
| Semantic     | 18.05% | Content quality, topic consistency            |
| Lineage      | 18.39% | Track record, authority, citation network     |

### TrustRank Formula

```
TrustRank = 300 + 550 * (0.1615 * P1 + 0.4741 * P2 + 0.1805 * P3 + 0.1839 * P4)
```

Score range: 300 (lowest) to 850 (highest).

### Trust Bands

| Band        | Range   | Meaning                                |
|-------------|---------|----------------------------------------|
| Exceptional | 800-850 | Top-tier trust, auto-approve for deals |
| Good        | 700-799 | Strong trust, recommended              |
| Established | 600-699 | Solid track record                     |
| Developing  | 500-599 | Growing, some gaps                     |
| At Risk     | 300-499 | Significant trust concerns             |

### Tier Classification

| Tier  | Followers   |
|-------|-------------|
| MEGA  | 10M+        |
| MACRO | 1M - 10M    |
| MID   | 100K - 1M   |
| MICRO | 10K - 100K  |
| NANO  | < 10K       |

## Snowflake UDF

If running inside Snowflake, call the TrustRank UDF directly:

```sql
SELECT suila_trustrank('mkbhd', 'youtube') AS trustrank_result;
```

Or batch-score creators:

```sql
SELECT handle, platform, suila_trustrank(handle, platform) AS score
FROM my_creator_list;
```

## MCP Server Configuration

Add to your MCP client config (Claude Desktop, Cursor, etc.):

```json
{
  "mcpServers": {
    "suilaai-creator-trustrank": {
      "url": "https://mcp.suilaai.com/creator-trustrank",
      "transport": "streamable-http",
      "description": "Score any content creator from 300-850 with 26 behavioral signals"
    }
  }
}
```

Once connected, agents can call `get_creator_trustrank` with a handle and platform.

## Example Agent Usage

An agent vetting creators for a brand campaign:

```
User: "Find me 5 tech reviewers on YouTube with TrustRank above 700"

Agent calls get_creator_trustrank for each candidate:
  mkbhd       → 736 (Good)
  linustechtips → 721 (Good)
  dave2d      → 708 (Good)
  unboxtherapy → 685 (Established) — below threshold
  ijustine    → 712 (Good)

Agent: "4 of 5 candidates scored above 700. UnboxTherapy scored 685
        due to lower engagement_quality (S10: 0.52). Recommending
        MKBHD, LinusTechTips, Dave2D, and iJustine."
```

## Rate Limits

- Public demo: 30 requests/minute per IP
- Authenticated API: Contact sales@suilaai.com
- Snowflake Native App: Unlimited (runs on your warehouse)

## Patent Notice

Protected under US Patent 12,505,169 B2 and Taiwan Patent I892115.