evaluate_influencerScore a creator on reach, fit, and risk.

Reach is easy. A million followers is just a number. What an agent actually needs is a defensible composite — reach, engagement quality, audience fit to a category, and brand-safety read of recent posts. evaluate_influencer returns all four as one object with reasons. Eight credits. Replaces the spreadsheet a junior team-member used to fill in.

When to call this tool

An agent should reach for it whenever a human is about to spend money on a creator partnership and needs a fast, repeatable read. Brand-fit before outreach. Risk-screen before contract. Sanity-check before a live campaign. The tool is opinionated about the four axes — that opinion is exactly what makes the score comparable across creators. It is not opinionated about the category; the agent passes a category string and the tool maps audience signal accordingly.

The brand-safety axis looks at the last 90 days of posts on the named platform and flags content that is hateful, deceptive, or off-brief for the category. The fit axis looks at audience composition signal — the kind of posts that get high engagement, the topics in the bio and pinned post — to estimate the share of the audience plausibly aligned with your category. None of this replaces a human judgement call; it just replaces the first 90 minutes of one.

Input schema

{
  "handle": "string  // required",
  "platform": "enum: 'x' | 'youtube' | 'instagram' | 'tiktok' | 'linkedin'",
  "category": "string  // for fit scoring",
  "include_recent_posts": "boolean  // default true"
}

Output schema

{
  "creator": { "handle", "display_name", "platform", "followers", "verified" },
  "reach_score": 0-100,
  "engagement_quality": 0-100,
  "fit_score": 0-100,
  "safety_score": 0-100,
  "composite": 0-100,
  "reasons": { "reach": "string", "engagement": "string", "fit": "string", "safety": "string" },
  "flags": [{ "type", "detail", "link" }],
  "credits_used": 8
}

Example invocations

1. Claude Desktop

Evaluate @swyx on X for a developer-tools partnership.
Show me reach, engagement, fit, and any safety flags.

2. ChatGPT custom GPT

When asked to "vet a creator", call evaluate_influencer.
Lead with the composite, then explain each axis.

3. Cursor MCP

Use evaluate_influencer to score these 8 YouTubers
for a B2B SaaS sponsorship. Output as a markdown table.

4. n8n

{ "jsonrpc": "2.0", "id": 1, "method": "tools/call",
  "params": { "name": "evaluate_influencer",
    "arguments": { "handle": "@thePrimeagen", "platform": "youtube", "category": "developer tools" } } }

5. Raw curl

curl -X POST https://www.mentionfox.com/mcp \
  -H "Authorization: Bearer $FOXAPIS_KEY" \
  -d '{"jsonrpc":"2.0","id":1,"method":"tools/call","params":{"name":"evaluate_influencer","arguments":{"handle":"@levelsio","platform":"x","category":"indie SaaS"}}}'

Sample output

{
  "creator": { "handle": "@levelsio", "followers": 580000, "verified": true },
  "reach_score": 82, "engagement_quality": 88,
  "fit_score": 94, "safety_score": 86,
  "composite": 87,
  "reasons": { "fit": "Audience is overwhelmingly indie founders and devs." },
  "credits_used": 8
}

Credit cost & rate limits

Flat 8 credits. 30 calls per minute. Each platform is read once; cross-platform creator reads should be done as separate calls and merged client-side.

Error codes & recovery