# MentionFox (remote · mentionfox.com)

Research on people and companies in your AI, every claim cited. A free account works.

- Trust score: 76/100 (medium)
- Registry status: active
- Liveness: live
- Owner verified: no
- Last scored: 2026-10-04

## Components

- remote · `mentionfox.com`: 76/100 (this document), [markdown](https://verifymcp.io/servers/osakasaul-mentionfox/mentionfox.md), [page](https://verifymcp.io/servers/osakasaul-mentionfox/mentionfox)

## Channel facts

- Endpoint: `https://mentionfox.com/mcp`
- Transports: `streamable-http`
- Auth: `none`
- Version: `1.0.1`

## Trust breakdown

How this component scores in each security and reliability category. Every signal is checked automatically against the live server, and we only credit what we can confirm. Scores are 0–100 per category. Scoring method: https://verifymcp.io/docs/scoring (what has changed: https://verifymcp.io/docs/scoring/changelog)

Scored 2026-10-04.

- **Endpoint Security**: 94/100
  - The endpoint's TLS certificate is valid, in date, and uses a strong key.
  - Authorisation is enforced on tool calls, advertised via RFC 9728 protected-resource metadata. Discovery is public, which costs nothing: no tool can be invoked without a token.
  - HTTPS is enforced; there's no plaintext access path.
  - The HSTS (Strict-Transport-Security) header is present.
  - DNSSEC check failed: this domain isn't protected by DNSSEC.
  - The authorisation server offers only Dynamic Client Registration (RFC 7591), which MCP 2026-07-28 deprecated in favour of Client ID Metadata Documents.
- **Transport & Reachability**: 100/100
  - Verified streamable-http transport via a live MCP handshake.
- **Schema Quality & AI Usability**: 59/100
  - AI-judged instruction clarity (excellent).
  - Context-footprint check failed: tool/resource definitions use about 2900 tokens (~322/item across 9 items; 9 tools + 0 resources), over budget; trim descriptions and params.
  - Usage-examples check failed: none of the tools include examples.
- **Stability & Change Management**: 17/100
  - Stability observed for 5 of 30 days with no destabilising changes; credit accrues until the full window elapses.
- **Tool Coverage**: 96/100
  - 100% of tools have a non-trivial description (not blank, and not just the tool's name).
  - 87% of tool parameters carry a description.
- **Tool Safety**: 100/100
  - No prompt-injection markers were found in the server instructions, tool names or descriptions we captured.
  - We read all 9 captured tool definition(s), and no name or description among them implies an irreversible operation.
  - An AI judge read all 10 captured unit(s) of tool text and found none that tries to manipulate the model reading it.
- **Capabilities**: 40/100
  - Spec-recency check failed: implements MCP spec 2025-03-26; the latest is 2026-07-28.

## Install

### How do I install the MentionFox MCP server?

MentionFox is a hosted endpoint at https://mentionfox.com/mcp, so there is nothing to install locally. Ready-made configuration for Claude, Cursor, VS Code, Codex and 5 more is on this page, copied from each client's own documentation.

### Claude

```bash
claude mcp add --transport http osakasaul-mentionfox 'https://mentionfox.com/mcp'
```

### Cursor

```json
{
  "mcpServers": {
    "osakasaul-mentionfox": {
      "url": "https://mentionfox.com/mcp"
    }
  }
}
```

### VS Code

```json
{
  "servers": {
    "osakasaul-mentionfox": {
      "type": "http",
      "url": "https://mentionfox.com/mcp"
    }
  }
}
```

### Codex

```toml
[mcp_servers.osakasaul-mentionfox]
url = "https://mentionfox.com/mcp"
```

### opencode

```json
{
  "$schema": "https://opencode.ai/config.json",
  "mcp": {
    "osakasaul-mentionfox": {
      "type": "remote",
      "url": "https://mentionfox.com/mcp",
      "enabled": true
    }
  }
}
```

### OpenClaw

```bash
openclaw mcp add osakasaul-mentionfox --url 'https://mentionfox.com/mcp' --transport streamable-http
```

### Hermes

```yaml
mcp_servers:
  osakasaul-mentionfox:
    url: "https://mentionfox.com/mcp"
```

### Netclaw

```json
{
  "McpServers": {
    "osakasaul-mentionfox": {
      "Transport": "http",
      "Url": "https://mentionfox.com/mcp"
    }
  }
}
```

### Vellum

```bash
assistant mcp add osakasaul-mentionfox -t streamable-http -u 'https://mentionfox.com/mcp'
```

### Other

```json
{
  "mcpServers": {
    "osakasaul-mentionfox": {
      "type": "http",
      "url": "https://mentionfox.com/mcp"
    }
  }
}
```

The mcpServers block is a cross-client convention. Remote transports vary, so check your client's docs.

## Changelog

Every change recorded for this component, newest first. Days that predate change tracking, or that we cannot explain, say so: "we were watching and nothing happened" and "we were not watching" are different claims.

### 2026-10-04 (score 76, +1)

- [security] Tool “get_full_report” rewrote its description, which is the text the model reads
- [cosmetic] “get_full_report” reworded the description of “company”
- [cosmetic] “get_full_report” reworded the description of “subject”

### 2026-10-02 (score 75, +1)

No change was recorded against any check on this day. Stability & Change Management went from 7 to 10. That category is still filling its 30-day observation window: 2 days of observed history at the previous scan, 3 at this one. The score rises as the window fills, whether or not the server changes.

### 2026-09-30 (score 74, 0)

- [security] The server rewrote its instructions, which are the text every model session reads
- [functional improvement] Stability: unverified → 0.03

### 2026-09-29 (score 74)

First indexed and scored.

## MCP tools (9)

### `source_check` (~396 tokens)

Someone says they are a doctor, a lawyer, a broker, a professor, a director of a company. This checks those particular claims against the official lists that would record them. Paste whatever you have — the pitch email, a name, a link — and the claims are read out of it. You get one of four answers: verified, partly verified, could not verify, or contradicted by the record. Under it: each claim that was confirmed, where it was confirmed and a link to the entry; each claim that was not, saying exactly which list was searched and came back empty; and a plain list of what was not checked at all, so a gap looks like a gap instead of a clean bill of health. It will never tell you somebody is fake. There is no such answer, because failing to find someone on a list is not evidence they are lying — plenty of real people are not on the list that would cover them, and "could not verify" means one search, on one day, in one place. What comes back is fixed at the moment it was made and never edited; checking again later produces a new one and leaves the first alone. Use it for "is this person who they say they are", "check this expert before I quote them", "should I trust this pitch", "is this doctor actually registered". Costs 25 credits and takes a few seconds.

Input parameters:

- `employer` (string): Optional. The organisation they say they are with, if it is not obvious from the input.
- `input` (string, required): The pasted pitch, name, email address or link naming the person to check. Paste it whole — the claims are read out of it.
- `name` (string): Optional. The subject's name, if it is not obvious from the input.
- `title` (string): Optional. The role or credential they state, if it is not obvious from the input.

### `enrich_person` (~212 tokens)

How to reach one person: their email address, phone number where there is one, and their profile links. Give a name — add their employer if the name is a common one — and you get back the person, their current role, and the ways to contact them. Use it when someone says "how do I get hold of her", "find his email", "I need to reach the person who runs marketing there". Do not use it to find out WHO to contact: this looks up somebody you can already name. If the name matches more than one person you get the list of candidates and nothing is charged; pick one and ask again. Nothing is charged when nobody matches either. Costs 5 credits when it finds them.

Input parameters:

- `company` (string): Company name or domain — disambiguation hint.
- `name` (string): Natural-language person name. Provide either name or person_id.
- `person_id` (string): UUID lock-in after disambiguation. Provide either name or person_id.

### `get_snapshot` (~515 tokens)

A short brief on one subject — a person, a firm, a publication — written from what is on the public record and returned as text plus a link to a formatted page you can share. Say who the subject is and what kind of subject they are, because the brief is written for the kind: what matters about a founder is not what matters about a doctor. What comes back is about a page: who they are, what the record shows, what stands out, and what to be careful of. Ask for the same subject twice and the second time is free — you are handed the brief already written, unchanged, rather than paying for it again. Nothing is charged when there is no such subject or too little on the record to say anything useful. Subject types you can ask for, and what each costs: company (5 credits), counterparty (5 credits), crypto_coin (5 credits), donor (5 credits), executive (5 credits), expert_witness (5 credits), founder (5 credits), insurance_underwriting (5 credits), investor (5 credits), journalism_person (5 credits), journalist (5 credits), litigation_funder (5 credits), nft_collection (5 credits), pe_firm (5 credits), penny_stock (5 credits), physician (10 credits), pre_ipo_secondary (5 credits), publication (5 credits), source (5 credits), spac_sponsor (5 credits), specific_fund (5 credits), vc_firm (5 credits), wealth_advisor (5 credits). Use this when someone says "who is this", "what should I know before I meet them", "quick read on this firm". For the long version, use get_full_report.

Input parameters:

- `company` (string): Optional. The organisation they are with, or the domain, when the name on its own could mean several people.
- `new_report` (boolean): Set this only when you want a fresh brief on a subject you already have one for. It writes and charges for a second brief and leaves the earlier one untouched. Leaving it off hands you the brief you…
- `subject` (string, required): Who or what the brief is about — a person's name, a firm, a publication. Add the employer or domain to it if the name alone is common.
- `subject_type` (string, required): Which kind of subject this is. The accepted values are listed in this tool's description, each with its price; they are read from the live price list, so ask for one of those exactly.

### `get_full_report` (~641 tokens)

The long version: a full written report on one subject, section by section, ending with the sources it drew on. Where the short brief runs to a page, this runs to several thousand words. It is fixed at the moment it is made — it records what the public record said on that day and is never edited afterwards, so what you show someone in six months is what you bought. If you want a fresher picture, you ask for a new one: it is written and paid for separately and the old one stays exactly as it was. Asking again for a subject you already have a report on hands you the one you already bought, free. Writing takes a few minutes; if it is still being written when the call comes back you get a link and can return to it. Costs 50 credits and needs the Pro plan. Subject types you can ask for: accelerator (50 credits), company (50 credits), counterparty (50 credits), crypto_coin (50 credits), donor (50 credits), executive (50 credits), expert_witness (50 credits), founder (50 credits), insurance_underwriting (50 credits), investor (50 credits), journalism_person (50 credits), journalist (50 credits), litigation_funder (50 credits), nft_collection (50 credits), pe_firm (50 credits), penny_stock (50 credits), physician (50 credits), pre_ipo_secondary (50 credits), publication (50 credits), source (50 credits), spac_sponsor (50 credits), specific_fund (50 credits), vc_firm (50 credits), wealth_advisor (50 credits). For a company, put its website in company (for example subject "Flock Safety", company "flocksafety.com") and the report starts on that company straight away; a website on its own as the subject works too. If you give only a name and more than one company goes by it, nothing is charged: you get a short list of the companies it could be, each with its website and a line on what it does, and you call again with the right website in company. Use this when someone says "I need the full write-up", "proper due diligence before we commit", "the long one".

Input parameters:

- `company` (string): Optional. The organisation they are with, or its website, when the name on its own could mean several. For a company report, put the company's website here (for example flocksafety.com) so the report…
- `new_report` (boolean): Set this only when you want a NEW report on a subject you already have one for. It writes and charges for a second report; the earlier one is left untouched. Leaving it off returns the report you alr…
- `subject` (string, required): Who or what the report is about — a person's name, a firm, a publication, or a company name or website.
- `subject_type` (string, required): Which kind of subject this is. The accepted values are listed in this tool's description; they are read from the live price list, so ask for one of those exactly.

### `get_dossier` (~291 tokens)

Everything on file about one person, in sections: what they have done, who they work with, how they are spoken about, what they have written or said in public, and what has been in the news about them lately. Broader than a short brief and not written for any one purpose — this is the file, not the verdict. Use it for "tell me everything about her", "what am I walking into with this person", "is this creator worth working with", "background on him before the call". If you only need an email, ask for their contact details instead — that is far cheaper. If a short written assessment is what you want, ask for a brief. If the name matches more than one person you get the candidates and nothing is charged; nothing is charged when nobody matches.Costs 30 credits the first time, and nothing at all if you have looked this person up before.

Input parameters:

- `company` (string): Company name — disambiguation hint when name is ambiguous.
- `format` (string): summary = the most actionable subset; full = every populated section
- `name` (string): Natural-language person name. Provide either name or person_id.
- `person_id` (string): UUID lock-in after disambiguation. Provide either name or person_id.
- `refresh` (boolean): Force a fresh enrichment instead of reading cache (charges 30 cr).

### `compare_people` (~226 tokens)

Two to five people set out side by side, line by line — how they work, how they come across, where their careers have gone, how deep their experience runs, how visible they are, who they are connected to. You get a table you can read down a column for one person or across a row to see how they differ. Tell it what the comparison is for and the framing follows: candidates for a role, people to pitch, names on a shortlist. It compares people you have already looked up; if fewer than two of them are on file it says so and charges nothing. If a name matches more than one person you get the candidates for that name, also free. Costs about 5 credits a head, so a three-way is around 15.

Input parameters:

- `context` (string): Optional context shaping the comparison framing (e.g. "investment targets", "speaking-spot candidates")
- `people` (array, required): Two to five people. Either plain names — ["Ada Lovelace", "Alan Turing"] — or objects carrying name or person_id.

### `scan_for_mentions` (~224 tokens)

What people said about something in the last day. Name a brand, a product, a topic or a person and this comes back with the posts, comments, articles and videos that mention it — each one with where it was said, a link to it, and whether it read as positive, negative or neither. It also marks the ones that sound like somebody about to buy, so this is the tool for "who is asking for something like ours". Use it for "what are people saying about us today", "did anyone pick up the launch", "anything new on this topic this week". It looks across 55 places by default and you can narrow that. It does not build a picture of a person — for that, ask for a brief. Costs 3 credits for each place it looks, so a default run of five places is about 15.

Input parameters:

- `hours_back` (integer)
- `sources` (array): Optional source filter (default = all 50+)
- `topic` (string, required): Topic, brand, or person to scan for

### `export_report` (~191 tokens)

Turn something you already have into a file you can send someone: a web page, a print-ready page you can save as a PDF, or plain text. Nothing about MentionFox appears on it, and you can put your own name, logo and colours on it, so it goes to a client as your work. Use it for "send me that as a PDF", "I need this as a document for the client", "put our branding on it". This makes a fixed copy of what exists now — for something that keeps up to date as the information changes, share the live link instead. Costs 5 credits.

Input parameters:

- `branding` (object): Optional white-label branding override
- `format` (string)
- `report_id` (string, required): For dossier: person_profile_id. For compare: comma-separated names. For others: the eval/report row id.
- `report_type` (string, required)

### `get_my_recent_research` (~113 tokens)

What you have already looked up. Lists the people you have researched, the write-ups you have had made, and the calls you have run here recently — yours only, nobody else's. Worth calling before you pay for something: if you looked this person up last month, you already have it. Use it for "have I checked this person before", "what did I look at last week", "show me what I have". Nothing is looked up and nothing is charged.

Input parameters:

- `limit` (integer)

## Diagnostics

Captured diagnostic sections: TLS, DNSSEC, Authorisation, Transports. The full working is on the page: https://verifymcp.io/servers/osakasaul-mentionfox/mentionfox#diagnostics

## Score history

- 2026-10-04: 76
- 2026-10-03: 75
- 2026-10-02: 75
- 2026-10-01: 74
- 2026-09-30: 74
- 2026-09-29: 74

## Common questions

### What is the MentionFox MCP server?

MentionFox is an MCP server listed in the public MCP registry as io.github.OsakaSaul/mentionfox. Research on people and companies in your AI, every claim cited. A free account works. This page covers its hosted endpoint (https://mentionfox.com/mcp).

### Is the MentionFox MCP server safe to use?

MentionFox scores 76 out of 100 on VerifyMCP. That is a record of what we were able to check automatically, not an endorsement. The category breakdown on this page shows every signal behind the number, including the ones we could not confirm.

### What tools does the MentionFox MCP server expose?

MentionFox exposes 9 tools: source_check, enrich_person, get_snapshot, get_full_report, get_dossier, and 4 more. Their descriptions and schemas cost roughly 2,809 tokens of context every time the server is loaded.

### Does the MentionFox MCP server require authentication?

Yes. MentionFox asked us for credentials when we connected, so you will need to authorise it in your MCP client before it can do anything.

### Is the MentionFox MCP server still maintained?

MentionFox is still listed as active in the MCP registry. We last reached this channel on 4 October 2026. Those dates come from our own scans of the registry and the channel itself, not from anything the publisher announced.

## Links

- Remote endpoint: https://mentionfox.com/mcp
- Website: https://mentionfox.com/?utm_source=mcp_registry&utm_medium=mcp_listing
- Changelog RSS feed: https://verifymcp.io/servers/osakasaul-mentionfox/mentionfox.xml
- Changelog JSON feed: https://verifymcp.io/servers/osakasaul-mentionfox/mentionfox.json
- HTML version of this page: https://verifymcp.io/servers/osakasaul-mentionfox/mentionfox