# `ai.theaggregate/the-aggregate` on Anlyon Radar

> Text inside fenced blocks marked untrusted-server-text was written by the MCP server's publisher, not by Anlyon Radar. 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://radar.anlyon.com/servers/ai.theaggregate/the-aggregate
- JSON: https://radar.anlyon.com/api/v1/servers/ai.theaggregate/the-aggregate
- Version: 1.0.2 (active)
- Publisher: ai.theaggregate (domain proven to the MCP Registry)
- Runs: remote (streamable-http)
- Sign-in to list tools: none
- Metadata from the official MCP Registry, synced 2026-10-21T17:43:40.029Z

Title and description, as the publisher wrote them:

```untrusted-server-text
The Aggregate — LLM benchmark aggregate
LLM rankings from public benchmarks, model comparisons and source-linked results. Updated daily.
```

Scanned 2026-11-11T22:18:17.983Z over streamable-http.

**8 tools: 1 change things and 7 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 | 7 | allow |

## Exfil surface

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

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

## Findings

No check matched anything in these tool definitions.


## Changes things (1)

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

### `about_the_aggregate`

- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.

Description, as the server wrote it:

```untrusted-server-text
What this data is: the published rating method, current coverage counts, update cadence, and how to cite it.
```

## Only reads (7)

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

### `compare_models`

- The name "compare_models" says "compare": a word for reading.
- Tag private-data: The description's first sentence says "note": it reads notes.

Description, as the server wrote it:

```untrusted-server-text
Head-to-head between 2-4 models: aggregate ranks, rating gap with a significance note based on the standard errors, and notable benchmarks they share.
```

### `get_benchmark`

- The name "get_benchmark" says "get": a word for reading.
- Tag untrusted-input: 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 benchmark in depth: what it measures, the original source leaderboard URL, IRT stats (difficulty, noise, model coverage), skill weights, and the current top models on it.
```

### `get_leaderboard`

- The name "get_leaderboard" says "get": a word for reading.

Description, as the server wrote it:

```untrusted-server-text
Top of the cross-benchmark aggregate ranking: published Aggregate Ratings combine public benchmark evidence on one scale (call about_the_aggregate for the current coverage and method). One row per model by default, fused across reasoning-effort settings. Supports paging via limit/offset.
```

### `get_model`

- The name "get_model" says "get": a word for reading.

Description, as the server wrote it:

```untrusted-server-text
One model in depth: aggregate rank, Aggregate Rating with standard error, provider, what it is, cost per task where known, and its most notable benchmark results (with percentiles).
```

### `get_prediction_duel`

- The name "get_prediction_duel" says "get": a word for reading.

Description, as the server wrote it:

```untrusted-server-text
Guesswork — the public prediction duel: every day frontier LLMs and The Aggregate's own IRT model predict newly scraped benchmark scores before seeing them, and the errors are scored. Returns the current monthly standings, wins and losses included.
```

### `search_benchmarks`

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

Description, as the server wrote it:

```untrusted-server-text
Find benchmarks in the aggregate by (partial) name. Returns model coverage, difficulty on the rating scale, and the benchmark page URL.
```

### `search_models`

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

Description, as the server wrote it:

```untrusted-server-text
Find ranked models by (partial) name or provider. Returns rank, Aggregate Rating and the model page URL. One row per model by default, fused across reasoning-effort settings.
```

## Claude Code permissions

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

```json
{
  "permissions": {
    "allow": [
      "mcp__the-aggregate__get_leaderboard",
      "mcp__the-aggregate__search_models",
      "mcp__the-aggregate__get_model",
      "mcp__the-aggregate__compare_models",
      "mcp__the-aggregate__search_benchmarks",
      "mcp__the-aggregate__get_benchmark",
      "mcp__the-aggregate__get_prediction_duel"
    ],
    "ask": [
      "mcp__the-aggregate__about_the_aggregate"
    ],
    "deny": []
  }
}
```

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

- `mcp__the-aggregate__compare_models`: private-data
- `mcp__the-aggregate__get_benchmark`: untrusted-input

## Remote posture

What an anonymous client saw at the remote endpoint, last 2026-11-20T18:28:14.006Z, unchanged since 2026-10-22T14:21:32.458Z. No credentials were sent and no tool was called. Values are shown as the endpoint returned them. Method: https://radar.anlyon.com/methodology#remote-posture

| Fact | Observed |
| --- | --- |
| Reachable | `yes` |
| HTTP status of an anonymous initialize | `200` |
| tools/list served without credentials | `yes` |
| Tools that change things, listed without credentials | `1` |
| Protocol version | `2025-06-18` |
| WWW-Authenticate on 401 | `not observed` |
| resource_metadata in WWW-Authenticate | `none` |
| RFC 9728 protected resource metadata | `none` |
| Authorization servers | `none` |
| Legacy HTTP+SSE transport | `no` |
| Dynamic client registration | `not observed` |
| Client ID metadata documents | `not observed` |

- **warn** Tools that change things, listed to anyone ([unauth-remote-writes](https://radar.anlyon.com/methodology#unauth-remote-writes))

```untrusted-server-text
The endpoint served tools/list without credentials, and the list has 1 tool that changes things (1 write). Whether a call also works without credentials was not tested: tools are never called.

https://theaggregate.ai/mcp: tools/list returned 8 tools to an anonymous client, including about_the_aggregate (write)
```
