# Incident Library (remote · incidentlib-mcp.benchlet.com)

Public post-incident reports as structured cases: search by symptom, read one, list by cause.

- Trust score: 78/100 (medium)
- Change this week: +4
- Registry status: active
- Liveness: live
- Owner verified: no
- Last scored: 2026-11-21

## Components

- remote · `incidentlib-mcp.benchlet.com`: 78/100 (this document), [markdown](https://verifymcp.io/servers/com-benchlet-incidentlib/incidentlib-mcp.md), [page](https://verifymcp.io/servers/com-benchlet-incidentlib/incidentlib-mcp)

## Channel facts

- Endpoint: `https://incidentlib-mcp.benchlet.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 2026-11-21.

- **Endpoint Security**: 83/100
  - The endpoint's TLS certificate is valid, in date, and uses a strong key.
  - No authorisation is required to call this server. Nothing on it is marked destructive, but we can't show that an anonymous caller can only read: full credit needs the server to list at least one tool and every tool to be verifiably read-only. We use a best-effort check of tool names and descriptions to tell read-only tools from tools that change things, so some genuinely read-only tools may not get full credit. Names that start with a read verb (get, list, read, search and similar) and accurate readOnlyHint annotations help. Meet that to earn full credit, or require authorisation.
  - HTTPS is enforced; there's no plaintext access path.
  - The HSTS (Strict-Transport-Security) header is present.
  - DNSSEC is configured correctly; the domain's records validate against the full chain to the root.
- **Transport & Reachability**: 100/100
  - Verified streamable-http transport via a live MCP handshake.
- **Schema Quality & AI Usability**: 65/100
  - AI-judged instruction clarity (excellent).
  - Context-footprint check failed: tool/resource definitions use about 1539 tokens (~307/item across 5 items; 5 tools + 0 resources), over budget; trim descriptions and params.
  - Usage-examples check failed: none of the tools include examples.
- **Stability & Change Management**: 37/100
  - Stability observed for 11 of 30 days with no destabilising changes; credit accrues until the full window elapses.
- **Tool Coverage**: 100/100
  - 100% of tools have a non-trivial description (not blank, and not just the tool's name).
  - 100% 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 5 captured tool definition(s), and no name or description among them implies an irreversible operation.
  - An AI judge read all 5 captured unit(s) of tool text and found none that tries to manipulate the model reading it.
- **Capabilities**: 100/100
  - Implements a supported MCP spec version (2025-11-25); the latest is 2026-07-28.

## Install

### How do I install the Incident Library MCP server?

Incident Library is a hosted endpoint at https://incidentlib-mcp.benchlet.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 com-benchlet-incidentlib 'https://incidentlib-mcp.benchlet.com/mcp'
```

### Cursor

```json
{
  "mcpServers": {
    "com-benchlet-incidentlib": {
      "url": "https://incidentlib-mcp.benchlet.com/mcp"
    }
  }
}
```

### VS Code

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

### Codex

```toml
[mcp_servers.com-benchlet-incidentlib]
url = "https://incidentlib-mcp.benchlet.com/mcp"
```

### opencode

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

### OpenClaw

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

### Hermes

```yaml
mcp_servers:
  com-benchlet-incidentlib:
    url: "https://incidentlib-mcp.benchlet.com/mcp"
```

### Netclaw

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

### Vellum

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

### Other

```json
{
  "mcpServers": {
    "com-benchlet-incidentlib": {
      "type": "http",
      "url": "https://incidentlib-mcp.benchlet.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.

### 2026-11-20 (score 78, +1)

No change was recorded against any check on this day. Stability & Change Management went from 30 to 33. That category is still filling its 30-day observation window: 9 days of observed history at the previous scan, 10 at this one. The score rises as the window fills, whether or not the server changes.

### 2026-11-18 (score 77, +1)

No change was recorded against any check on this day. Stability & Change Management went from 23 to 27. That category is still filling its 30-day observation window: 7 days of observed history at the previous scan, 8 at this one. The score rises as the window fills, whether or not the server changes.

### 2026-11-16 (score 76, +2)

- [security improvement] DNSSEC: fail → pass

### 2026-11-15 (score 74, 0)

- [security] Tool “search_incidents” rewrote its description, which is the text the model reads

### 2026-11-14 (score 74, +1)

No change was recorded against any check on this day. Stability & Change Management went from 10 to 13. That category is still filling its 30-day observation window: 3 days of observed history at the previous scan, 4 at this one. The score rises as the window fills, whether or not the server changes.

### 2026-11-12 (score 73, +1)

- [security] Tool “get_incident” rewrote its description, which is the text the model reads
- [security] Tool “get_topic” rewrote its description, which is the text the model reads
- [security] Tool “list_incidents” rewrote its description, which is the text the model reads
- [security] Tool “vocab” rewrote its description, which is the text the model reads
- [functional regression] Schema quality: 213 → 298

### 2026-11-11 (score 72, +7)

- [security improvement] Authorization: unverified → partial
- [functional improvement] Tool coverage: 93% → 100%
- [functional improvement] Stability: unverified → 0.03
- [functional] Schema quality: good → excellent
- [cosmetic] “get_topic” reworded the description of “kind”
- [cosmetic] “vocab” reworded the description of “limit”

### 2026-11-10 (score 65)

First indexed and scored.

## MCP tools (5)

### `search_incidents` (~492 tokens)

Search incidents by symptoms

Use first, whenever someone describes a failure they are seeing. Pass the symptoms as a sentence or a few words, in English or Korean; returns the reviewed public post-incident reports that answer it, best match first: the search ranks every case and keeps only those judged relevant to the sentence, usually a handful, and none when the library has no such incident (`total` 0). `total` is how many, `results` is one page of them in full (each with the case id, the page url and the source link), `matched` is every one as a row, and `next_cursor` fetches the next page. `interpreted` lists the values the sentence named (the search moves cases carrying them up) and any organisation it named (a filter to choose, not a ranking signal); `facets` count the results by root cause, component, symptom, contributing factor and organisation with the case ids, which answers 'what usually caused this' from the same call. The optional slug filters keep only cases carrying those values (AND); take slugs from facets, interpreted or vocab. Follow with get_incident for the full case.

Input parameters:

- `component` (string): Keep only cases with this component slug, e.g. coredns
- `cursor` (string): next_cursor from the previous page
- `domain` (string): Keep only cases in this domain, e.g. kubernetes or ai-infra
- `factor` (string): Keep only cases with this contributing factor slug, e.g. no-alerting
- `lang` (string): ko returns the Korean text of each field where the reviewers wrote one
- `limit` (integer): Cases per page
- `org` (string): Keep only cases of this organisation, by slug, e.g. github (from facets.orgs, interpreted or vocab)
- `root_cause` (string): Keep only cases with this root cause slug, e.g. capacity
- `symptom` (string): Keep only cases with this symptom slug, e.g. pod-pending
- `symptoms` (string): What is observed, as a sentence or a few words, e.g. 'pods cannot get an IP, nodes show no available addresses' or 'coredns oom'. Leave it out to list every case under the slug filters, newest first;…

### `get_incident` (~209 tokens)

Get one incident

Returns one case in full by its id: the structured write-up (symptoms, trigger, root cause, components, contributing factors, summary, impact, timeline, detection, remediation, action items, lessons), its sources with archived copies, and `similar_cases`. Use after search_incidents, list_incidents or get_topic has given you a case id; not for finding cases (use search_incidents for symptoms, list_incidents or get_topic for a category). `lang: ko` returns the Korean text of each field where the reviewers wrote one, otherwise English. The id must look like IL-0007; an id of that shape that matches no approved case returns an error result saying so, and an id of another shape is rejected before the call. Every call for the same id returns the same case.

Input parameters:

- `id` (string, required): Case id such as IL-0007
- `lang` (string): ko returns the Korean text of each field where the reviewers wrote one

### `list_incidents` (~344 tokens)

List incidents by vocabulary value

Lists cases by category, newest first, one page at a time: `count` and `cases` (id, title, organisation, date, root cause, page url), plus `next_cursor` when more remain; call again with `cursor` for the following page. Use when the question names a category (a root cause, component, symptom, contributing factor, domain or organisation) rather than a symptom sentence; for a sentence use search_incidents, and for one category with its related categories counted use get_topic. Each filter is a slug (kebab-case) of its own kind: `symptom` is what was observed (pod-pending), `factor` is a condition that made it worse (no-alerting), `root_cause` is why it happened (capacity). Filters combine with AND. An unknown slug returns count 0 with no error; find the slug with vocab when you only have words. Reads only; the same filters return the same page.

Input parameters:

- `component` (string): e.g. coredns, vpc-cni
- `cursor` (string): next_cursor from the previous page
- `domain` (string): kubernetes or ai-infra
- `factor` (string): contributing factor, e.g. no-alerting
- `lang` (string): ko returns the Korean text of each field where the reviewers wrote one
- `limit` (integer): Cases per page
- `org` (string): the organisation's slug, e.g. github
- `root_cause` (string): e.g. capacity, config-change
- `symptom` (string): e.g. pod-pending

### `get_topic` (~289 tokens)

Get a root cause, component, symptom, contributing factor or organisation

Returns one vocabulary value with every case that carries it and, under `related`, the other four kinds counted over those cases with case ids: for a symptom, the root causes behind it (answers 'what usually causes this'); for a root cause, the components and factors seen with it. Use when you already have a slug and want the picture around it; use vocab to find the slug from words, list_incidents for a plain paged list, search_incidents for a symptom sentence. `kind` is the slug's category: root_causes, components, symptoms, factors or orgs (the organisations that published cases). With `symptoms`, the page keeps only the cases that sentence finds, best match first, and `query` echoes it, so 'certificate expiry, but only where DNS was involved' is one call. The answer is not paged: `count` and `cases` hold every case of the value (at most a few dozen today). An unknown slug or a slug of the wrong kind returns an error result naming it; nothing is written.

Input parameters:

- `kind` (string, required): Which category the slug belongs to
- `slug` (string, required): The value's slug, e.g. capacity, coredns, pod-pending, no-alerting, github
- `symptoms` (string): A symptom sentence that restricts the page to the cases it finds

### `vocab` (~205 tokens)

Vocabulary: slugs and their names

Resolves words to vocabulary slugs for the filters of search_incidents, list_incidents and get_topic. With `query`, returns `matches`: the `limit` values nearest in meaning to the words, nearest first, each with kind, slug, English and Korean name and page url (e.g. 'cert ran out' finds certificate-expiry); `kind` restricts to one category. There is no similarity cutoff: the list is always `limit` long, so when the words name nothing in the library the matches are far from them; read the names before using a slug, and when none fits, search with search_incidents instead of filtering. Without `query`, returns every value of every kind with names, aliases and definitions (a large answer; prefer `query`). Reads only.

Input parameters:

- `kind` (string): Only this kind
- `limit` (integer): Values to return
- `query` (string): A few words naming what you are looking for

## Diagnostics

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

## Score history

- 2026-11-21: 78
- 2026-11-20: 78
- 2026-11-19: 77
- 2026-11-18: 77
- 2026-11-17: 76
- 2026-11-16: 76
- 2026-11-15: 74
- 2026-11-14: 74
- 2026-11-13: 73
- 2026-11-12: 73
- 2026-11-11: 72
- 2026-11-10: 65

## Common questions

### What is the Incident Library MCP server?

Incident Library is an MCP server listed in the public MCP registry as com.benchlet/incidentlib. Public post-incident reports as structured cases: search by symptom, read one, list by cause. This page covers its hosted endpoint (https://incidentlib-mcp.benchlet.com/mcp).

### Is the Incident Library MCP server safe to use?

Incident Library scores 78 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 Incident Library MCP server expose?

Incident Library exposes 5 tools: search_incidents, get_incident, list_incidents, get_topic, vocab. Their descriptions and schemas cost roughly 1,539 tokens of context every time the server is loaded.

### Does the Incident Library MCP server require authentication?

No. We connected to Incident Library without credentials and it answered, so anything it exposes is reachable by anyone who knows the address.

### Is the Incident Library MCP server still maintained?

Incident Library is still listed as active in the MCP registry. We last reached this channel on 21 November 2026. Those dates come from our own scans of the registry and the channel itself, not from anything the publisher announced.

### What did VerifyMCP test on the Incident Library MCP server?

VerifyMCP ran 13 checks on Incident Library, covering TLS certificate, Authorization, HTTPS, HSTS header and 9 more. Supply Chain Security and Provenance & Transparency do not apply to a hosted endpoint, so they were not tested. The score does not measure everything, and the scoring documentation lists what it leaves out.

## Links

- Remote endpoint: https://incidentlib-mcp.benchlet.com/mcp
- Repository: https://github.com/Team-DogFoot/incidentlib-reports
- Website: https://benchlet.com/incidentlib/
- Changelog RSS feed: https://verifymcp.io/servers/com-benchlet-incidentlib/incidentlib-mcp.xml
- Changelog JSON feed: https://verifymcp.io/servers/com-benchlet-incidentlib/incidentlib-mcp/changelog.json
- Trust report (JSON): https://verifymcp.io/servers/com-benchlet-incidentlib/incidentlib-mcp.json
- HTML version of this page: https://verifymcp.io/servers/com-benchlet-incidentlib/incidentlib-mcp
