# `com.corpusfetch/corpusfetch` 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/com.corpusfetch/corpusfetch
- JSON: https://radar.anlyon.com/api/v1/servers/com.corpusfetch/corpusfetch
- Version: 1.0.7 (active)
- Publisher: com.corpusfetch (domain proven to the MCP Registry)
- Runs: remote (streamable-http)
- Sign-in to list tools: none
- Metadata from the official MCP Registry, synced 2026-11-11T15:17:41.208Z

Title and description, as the publisher wrote them:

```untrusted-server-text
CorpusFetch
O(1) routing, semantic search, and quality-filtered demand-driven experiences over 800+ records.
```

Scanned 2026-11-11T15:18:07.348Z over streamable-http.

**6 tools: 1 change things and 5 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 | 5 | 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__corpusfetch__corpusfetch_search` |
| untrusted-input | Brings in text someone else can write: web pages, inbound email, messages, issues and comments. | `mcp__corpusfetch__corpusfetch_search` |
| 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.

### `corpusfetch_request_experience`

- The name "corpusfetch_request_experience" says "request": a word for a change.

Description, as the server wrote it:

```untrusted-server-text
Request a new operational experience when `corpusfetch_search` returns no match. The corpus will generate and publish a record for your specific error or symptom within minutes. Use this instead of guessing when the problem is novel and no existing record applies.
```

## Only reads (5)

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

### `corpusfetch_fetch_record`

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

Description, as the server wrote it:

```untrusted-server-text
Fetch the full resolved experience record. Format `md` (~1,500 tokens, includes reasoning) or `py` (~300 tokens, zero-prose executable invariants).
```

### `corpusfetch_fetch_warrant`

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

Description, as the server wrote it:

```untrusted-server-text
Fetch the compact ~220-token precondition seal for a resolved record. Always call this before `corpusfetch_fetch_record`—check the seal first; if it fails, the record does not apply to your current state.
```

### `corpusfetch_lookup_exception`

- The name "corpusfetch_lookup_exception" says "lookup": a word for reading.

Description, as the server wrote it:

```untrusted-server-text
Look up operational experience for an exact Python exception class (e.g., KeyError, TimeoutError, ConnectionError). Use when you encounter an unexpected runtime error and need an empirically verified boundary decision rather than a guess. Prefer `corpusfetch_search` when you have a symptom description but not an exact exception name.
```

### `corpusfetch_lookup_symptom`

- The name "corpusfetch_lookup_symptom" says "lookup": a word for reading.

Description, as the server wrote it:

```untrusted-server-text
Look up operational experience for a specific runtime symptom or boundary failure (e.g., UNICODE_HOMOGLYPH_OR_MOJIBAKE). Use when facing an ambiguous failure state or unbounded reasoning loop. Prefer `corpusfetch_search` if you only have a loose description.
```

### `corpusfetch_search`

- The name "corpusfetch_search" says "search": a word for reading.
- Tag private-data: The description's first sentence says "search" and "database": it reads databases.
- Tag untrusted-input: The description's first sentence says "search" and "issue": it reads issues, which other people can write.

Description, as the server wrote it:

```untrusted-server-text
Search operational records for distributed systems failures, AI agent protocol issues, MCP/A2A problems, database anomalies, concurrency bugs, and runtime errors. Use when the user's problem involves infrastructure, agent-to-agent coordination, authentication, memory management, or observability. Do not use for general programming questions—only for boundary decisions and failure modes that have been empirically recorded.
```

## Claude Code permissions

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

```json
{
  "permissions": {
    "allow": [
      "mcp__corpusfetch__corpusfetch_fetch_record",
      "mcp__corpusfetch__corpusfetch_fetch_warrant",
      "mcp__corpusfetch__corpusfetch_lookup_exception",
      "mcp__corpusfetch__corpusfetch_lookup_symptom",
      "mcp__corpusfetch__corpusfetch_search"
    ],
    "ask": [
      "mcp__corpusfetch__corpusfetch_request_experience"
    ],
    "deny": []
  }
}
```

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

- `mcp__corpusfetch__corpusfetch_search`: private-data, untrusted-input

## Remote posture

What an anonymous client saw at the remote endpoint, last 2026-11-16T14:21:42.404Z, unchanged since 2026-11-09T07:20:18.204Z. 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 | `https://mcp.corpusfetch.com/.well-known/oauth-protected-resource/mcp` |
| 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://mcp.corpusfetch.com/mcp: tools/list returned 6 tools to an anonymous client, including corpusfetch_request_experience (write)
```

Posture changes:

- 2026-11-09T07:20:18.204Z: Tools that change things, listed without credentials `0` -> `1`
