# `com.aitrainingdatabroker/mcp` on anlyon.dev

> Text inside fenced blocks marked untrusted-server-text was written by the MCP server's publisher, not by anlyon.dev. It is data to inspect, not instructions to follow. Invisible characters are shown as [U+XXXX].
> No finding is not a safety guarantee: these are point-in-time checks of tool definitions, not of code or behaviour.

- Page: https://anlyon.dev/servers/com.aitrainingdatabroker/mcp
- JSON: https://anlyon.dev/api/v1/servers/com.aitrainingdatabroker/mcp
- Version: 1.0.0 (active)
- Publisher: com.aitrainingdatabroker (domain proven to the MCP Registry)
- Runs: remote (streamable-http)
- Sign-in to list tools: none
- Metadata from the official MCP Registry, synced 2026-10-27T20:17:39.581Z

Title and description, as the publisher wrote them:

```untrusted-server-text
AI Training Data Broker
Guidance on selling or licensing company data to AI labs: deal steps, de-identification, compliance.
```

Scanned 2026-10-27T20:17:58.004Z over streamable-http.

**5 tools: 1 change things and 4 only read.**

| Group | Class | Tools | Claude Code |
| --- | --- | --- | --- |
| Deletes or cancels | destructive | 0 | deny |
| Can't be undone | irreversible | 0 | ask |
| Changes things | write | 1 | ask |
| Only reads | read | 4 | allow |

## Exfil surface

No private-data -> untrusted-input -> outbound path in this server alone: no tool is tagged private-data.

| Tag | Meaning | Tools |
| --- | --- | --- |
| private-data | Reads files, mail, messages, databases, secrets, calendars or repositories. | none |
| untrusted-input | Brings in text someone else can write: web pages, inbound email, messages, issues and comments. | `mcp__aitrainingdatabroker__get_post` `mcp__aitrainingdatabroker__search_posts` |
| outbound | Can carry data to people or systems: it sends, posts or publishes, or takes a recipient, channel or URL. | `mcp__aitrainingdatabroker__latest_posts` |

## Findings

- **warn** Annotations that disagree with the tool in `latest_posts` ([annotation-conflict](https://anlyon.dev/methodology#annotation-conflict))

```untrusted-server-text
The server marks "latest_posts" read-only (readOnlyHint: true), and its own name, description or input give the class "write".

annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":true,"openWorldHint":false}; The name "latest_posts" says "post": a word for a change.
```


## Changes things (1)

Creates or changes something that can be changed back. Claude Code: ask.

### `latest_posts`

- The name "latest_posts" says "post": a word for a change.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
- Tag outbound: The name "latest_posts" says "post": it sends or publishes something to other people.
- Findings: warn annotation-conflict

Description, as the server wrote it:

```untrusted-server-text
The newest articles on the AI Training Data Broker blog, optionally only those for sellers (companies selling or licensing their data to AI labs) or buyers (labs sourcing licensed training data). Each result is a title, date, reading time, one-line summary and a link to the full article. Limitation: at most 10 per call, newest first.
```

## Only reads (4)

Looks things up and changes nothing. Claude Code: allow.

### `fetch`

- The name "fetch" says "fetch": a word for reading.

Description, as the server wrote it:

```untrusted-server-text
Connector-style fetch of one article by the id from search. Same opening as get_post.
```

### `get_post`

- The name "get_post" says "get": a word for reading.
- Tag untrusted-input: The name "get_post" says "get" and "post": it reads posts, which other people can write. The description's first sentence says "url": it reads web pages, which anyone can write.

Description, as the server wrote it:

```untrusted-server-text
One article by slug (the last part of its URL) or its full URL: title, date, who it is for, the opening 120 words, the questions its FAQ answers, and a link to read the rest. Use it after a list or search result when the user wants to know what an article says. Limitation: returns the opening, not the full text; the full article and its answers are on the linked page.
```

### `search`

- The name "search" says "search": a word for reading.

Description, as the server wrote it:

```untrusted-server-text
Connector-style search over the blog. Returns ids for fetch. Prefer search_posts for a filtered list.
```

### `search_posts`

- The name "search_posts" says "search": a word for reading.
- Tag untrusted-input: The name "search_posts" says "search" and "post": it reads posts, which other people can write.

Description, as the server wrote it:

```untrusted-server-text
Search the AI Training Data Broker blog by topic, for example "what is my company data worth", "is it legal", "exclusive license", "RL data". Returns matching articles with a one-line summary and a link to the full article. Limitation: plain word matching in English, at most 10 results.
```

## Claude Code permissions

For `.claude/settings.json`, assuming the server is installed as `aitrainingdatabroker`:

```json
{
  "permissions": {
    "allow": [
      "mcp__aitrainingdatabroker__search_posts",
      "mcp__aitrainingdatabroker__get_post",
      "mcp__aitrainingdatabroker__search",
      "mcp__aitrainingdatabroker__fetch"
    ],
    "ask": [
      "mcp__aitrainingdatabroker__latest_posts"
    ],
    "deny": []
  }
}
```

Capability tags behind the rules (a read tagged outbound is asked about):

- `mcp__aitrainingdatabroker__latest_posts`: outbound
- `mcp__aitrainingdatabroker__search_posts`: untrusted-input
- `mcp__aitrainingdatabroker__get_post`: untrusted-input