# `au.com.horsetruth/machine-intelligence` 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/au.com.horsetruth/machine-intelligence
- JSON: https://anlyon.dev/api/v1/servers/au.com.horsetruth/machine-intelligence
- Version: 1.0.0 (active)
- Publisher: au.com.horsetruth (domain proven to the MCP Registry)
- Runs: remote (streamable-http)
- Sign-in to list tools: none
- Metadata from the official MCP Registry, synced 2026-10-21T17:45:49.728Z

Title and description, as the publisher wrote them:

```untrusted-server-text
Horse Truth Machine Intelligence
Derived Australian racehorse intelligence for AI agents, MCP clients, software and publishers.
```

Scanned 2026-10-21T18:43:51.761Z over streamable-http.

**11 tools: 1 do something that can't be undone, 6 change things and 4 only read.**

| Group | Class | Tools | Claude Code |
| --- | --- | --- | --- |
| Deletes or cancels | destructive | 0 | deny |
| Can't be undone | irreversible | 1 | ask |
| Changes things | write | 6 | ask |
| Only reads | read | 4 | allow |

## Findings

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

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

annotations: {"readOnlyHint":true,"destructiveHint":false,"openWorldHint":false}; The name "discover_purchase_options" says "purchase": a word that usually means money moves, which cannot be taken back.
```


## Can't be undone (1)

Sends, pays, publishes or merges: once done, it cannot be taken back. Claude Code: ask.

### `discover_purchase_options`

- The name "discover_purchase_options" says "purchase": a word that usually means money moves, which cannot be taken back.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
- Findings: warn annotation-conflict

Description, as the server wrote it:

```untrusted-server-text
FREE: Return the self-service purchase, products, sandbox and checkout navigation needed for an autonomous client to buy Horse Truth machine access. Use this tool when you need to know how to pay: x402 per call, a direct credit pack, or the Apify marketplace. Use sandbox_preview instead to see what a result looks like.
```

## Changes things (6)

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

### `horse_changes`

- 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
PAID: US$0.02 per successful autonomous x402 call, or 1 Horse Truth direct credit. Return material Horse Truth change events for one horse. Use this tool when you want the history of material changes for one horse, newest first, each with the fields that changed and when. Use horse_snapshot instead for the current state, and horse_explanation for the reasons behind it.
```

### `horse_explanation`

- 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
PAID: US$0.02 per successful autonomous x402 call, or 1 Horse Truth direct credit. Return a structured explanation of the current derived Horse Truth read without exposing raw provider records. Use this tool when you want the reasons behind the current read, as a list of factors each with its derived value. Use horse_snapshot instead for the summary values themselves, and horse_changes for what changed over time.
```

### `horse_intelligence`

- 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
PAID: US$0.02 per successful autonomous x402 call, or 1 Horse Truth direct credit. Return derived intelligence for one horse. Raw provider records are not exposed. Use this tool when you need the full derived state of one horse, including race history and every read. Use horse_snapshot instead for a short summary, horse_explanation for the reasons only, and horse_changes for what changed.
```

### `horse_provenance`

- 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
PAID: US$0.02 per successful autonomous x402 call, or 1 Horse Truth direct credit. Return a cryptographic receipt for the current derived horse-intelligence payload. Use this tool when you need a cryptographic receipt for the current read, so you can later prove what was returned. Use horse_intelligence instead for the read itself.
```

### `horse_rankings`

- 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
PAID: US$0.02 per successful autonomous x402 call, or 1 Horse Truth direct credit. Return current derived Horse Truth rankings from the already-warmed canonical model cache. Use this tool when you want the current ranked list of horses in the Horse Truth model. Use horse_snapshot or horse_compare instead for named horses.
```

### `horse_signal`

- 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
PAID: US$0.02 per successful autonomous x402 call, or 1 Horse Truth direct credit. Return one named derived Horse Truth signal for a horse, such as readiness, biomechanics, progression, trouble watch, class, reliability, alerts or winning conditions. signal=race_day_signals returns every race-day signal Horse Truth measured for the horse (ran better than its finish, beat the day's track pattern, running profile, form turning point). Use this tool when you need one named signal only (for example readiness or class_rating). Use horse_snapshot instead for several signals at once, and horse_intelligence for everything. signal=stride_measures returns the horse's stride measures from race tracking data: held its stride late, lifted its stride rate late, stride against its own runs and stride type.
```

## Only reads (4)

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

### `horse_compare`

- The name "horse_compare" says "compare": a word for reading.

Description, as the server wrote it:

```untrusted-server-text
PAID: US$0.02 per successful autonomous x402 call, or 1 Horse Truth direct credit. Compare compact derived snapshots for two to five horses. Use this tool when you want two to five horses side by side on the same derived measures. Use horse_snapshot instead for a single horse.
```

### `horse_snapshot`

- The name "horse_snapshot" says "snapshot": a word for reading.

Description, as the server wrote it:

```untrusted-server-text
PAID: US$0.02 per successful autonomous x402 call, or 1 Horse Truth direct credit. Return a compact derived operating snapshot for one horse. Use this tool when you want a short current summary of one horse: readiness, progression, soundness and trouble watch, closing par, latest run and alerts. Use horse_intelligence instead for the full state, horse_explanation for the reasons behind the read, and horse_changes for what changed over time.
```

### `resolve_horse`

- The name "resolve_horse" says "resolve": a word for reading.

Description, as the server wrote it:

```untrusted-server-text
PAID: US$0.001 per successful autonomous x402 call, or 1 Horse Truth direct credit. Resolve a horse identity against Horse Truth profiles. Use this tool when you have a partial or uncertain horse name and need the exact Horse Truth identity before other calls. Use horse_snapshot or horse_intelligence instead once you know the exact name.
```

### `sandbox_preview`

- The name "sandbox_preview" says "preview": a word for reading.

Description, as the server wrote it:

```untrusted-server-text
FREE: Return a real derived Horse Truth proof for today's default sample horse (picked once a day by a published rule, with a deep, stable read) plus the cheapest exact paid next actions. Pass an optional horse name to get the free public teaser for that horse instead. No paid entitlement is consumed and no raw provider records are exposed. Use this tool when you want to see a real example result, or the free public teaser for one horse, before paying. Use discover_purchase_options instead to learn how to pay, and horse_intelligence for the full paid read.
```

## Claude Code permissions

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

```json
{
  "permissions": {
    "allow": [
      "mcp__machine-intelligence__horse_compare",
      "mcp__machine-intelligence__horse_snapshot",
      "mcp__machine-intelligence__resolve_horse",
      "mcp__machine-intelligence__sandbox_preview"
    ],
    "ask": [
      "mcp__machine-intelligence__discover_purchase_options",
      "mcp__machine-intelligence__horse_changes",
      "mcp__machine-intelligence__horse_explanation",
      "mcp__machine-intelligence__horse_intelligence",
      "mcp__machine-intelligence__horse_provenance",
      "mcp__machine-intelligence__horse_rankings",
      "mcp__machine-intelligence__horse_signal"
    ],
    "deny": []
  }
}
```