# Agent Reliability (remote · agentreliability.dev)

Testing, benchmarking and auditing autonomous AI agents — methods, harnesses, evidence

- Trust score: 89/100 (high trust)
- Change this week: +3
- Registry status: active
- Liveness: live
- Owner verified: no
- Last scored: 2026-09-20

## Components

- remote · `agentreliability.dev`: 89/100 (this document), [markdown](https://verifymcp.io/servers/dev-agentreliability-agent-reliability/agentreliability.md), [page](https://verifymcp.io/servers/dev-agentreliability-agent-reliability/agentreliability)

## Channel facts

- Endpoint: `https://agentreliability.dev/mcp?via=manifest`
- Transports: `streamable-http`
- Auth: `none`
- Version: `0.6.14`

## 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 2026-09-20.

- **Endpoint Security**: 80/100
  - The endpoint's TLS certificate is valid, in date, and uses a strong key.
  - No authorisation is required to call this server. Every tool declares its destructiveHint and none is destructive, so open access doesn't expose one.
  - HTTPS is enforced; there's no plaintext access path.
  - The HSTS (Strict-Transport-Security) header is present.
  - DNSSEC check failed: this domain isn't protected by DNSSEC.
- **Transport & Reachability**: 100/100
  - Verified streamable-http transport via a live MCP handshake.
- **Schema Quality & AI Usability**: 91/100
  - 100% of prompts and resources have a non-trivial description (not blank, and not just the item's name).
  - AI-judged instruction clarity (excellent).
  - Tool/resource definitions use about 4561 tokens (~89/item across 51 items; 9 tools + 42 resources), lean.
  - Usage-examples check failed: none of the tools include examples.
- **Stability & Change Management**: 93/100
  - Stability observed for 28 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.
  - Structured output schemas are declared (100% of tools); any adoption earns full credit.
- **Tool Safety**: 100/100
  - No prompt-injection markers were found in the server instructions, tool names or descriptions we captured.
  - We read all 9 captured tool definition(s), and no name or description among them implies an irreversible operation.
  - An AI judge read all 10 captured unit(s) of tool text and found none that tries to manipulate the model reading it.
- **Capabilities**: 100/100
  - Implements a current MCP spec version (2026-07-28).

## Install

### How do I install the Agent Reliability MCP server?

Agent Reliability is a hosted endpoint at https://agentreliability.dev/mcp?via=manifest, 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 dev-agentreliability-agent-reliability 'https://agentreliability.dev/mcp?via=manifest'
```

### Cursor

```json
{
  "mcpServers": {
    "dev-agentreliability-agent-reliability": {
      "url": "https://agentreliability.dev/mcp?via=manifest"
    }
  }
}
```

### VS Code

```json
{
  "servers": {
    "dev-agentreliability-agent-reliability": {
      "type": "http",
      "url": "https://agentreliability.dev/mcp?via=manifest"
    }
  }
}
```

### Codex

```toml
[mcp_servers.dev-agentreliability-agent-reliability]
url = "https://agentreliability.dev/mcp?via=manifest"
```

### opencode

```json
{
  "$schema": "https://opencode.ai/config.json",
  "mcp": {
    "dev-agentreliability-agent-reliability": {
      "type": "remote",
      "url": "https://agentreliability.dev/mcp?via=manifest",
      "enabled": true
    }
  }
}
```

### OpenClaw

```bash
openclaw mcp add dev-agentreliability-agent-reliability --url 'https://agentreliability.dev/mcp?via=manifest' --transport streamable-http
```

### Hermes

```yaml
mcp_servers:
  dev-agentreliability-agent-reliability:
    url: "https://agentreliability.dev/mcp?via=manifest"
```

### Netclaw

```json
{
  "McpServers": {
    "dev-agentreliability-agent-reliability": {
      "Transport": "http",
      "Url": "https://agentreliability.dev/mcp?via=manifest"
    }
  }
}
```

### Vellum

```bash
assistant mcp add dev-agentreliability-agent-reliability -t streamable-http -u 'https://agentreliability.dev/mcp?via=manifest'
```

### Other

```json
{
  "mcpServers": {
    "dev-agentreliability-agent-reliability": {
      "type": "http",
      "url": "https://agentreliability.dev/mcp?via=manifest"
    }
  }
}
```

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.

### 2026-09-18 (score 89, +1)

No change was recorded against any check on this day. Stability & Change Management went from 83 to 87. That category is still filling its 30-day observation window: 25 days of observed history at the previous scan, 26 at this one. The score rises as the window fills, whether or not the server changes.

### 2026-09-16 (score 88, +1)

No change was recorded against any check on this day. Stability & Change Management went from 77 to 80. That category is still filling its 30-day observation window: 23 days of observed history at the previous scan, 24 at this one. The score rises as the window fills, whether or not the server changes.

### 2026-09-15 (score 87, 0)

- [functional] Server version: 0.9.0 → 0.9.2

### 2026-09-14 (score 87, +1)

No change was recorded against any check on this day. Stability & Change Management went from 70 to 73. That category is still filling its 30-day observation window: 21 days of observed history at the previous scan, 22 at this one. The score rises as the window fills, whether or not the server changes.

### 2026-09-12 (score 86, +1)

No change was recorded against any check on this day. Stability & Change Management went from 63 to 67. That category is still filling its 30-day observation window: 19 days of observed history at the previous scan, 20 at this one. The score rises as the window fills, whether or not the server changes.

### 2026-09-10 (score 85, +1)

No change was recorded against any check on this day. Stability & Change Management went from 57 to 60. That category is still filling its 30-day observation window: 17 days of observed history at the previous scan, 18 at this one. The score rises as the window fills, whether or not the server changes.

### 2026-09-09 (score 84, +59)

- [security improvement] Authorization: unverified → partial
- [security improvement] Injection markers: unverified → pass
- [security improvement] Transport: fail → pass
- [security] Tool “get_related” rewrote its description, which is the text the model reads
- [functional improvement] Endpoint reachability: not serving MCP → reachable
- [functional improvement] Tool coverage: unverified → 100
- [functional improvement] Schema quality: unverified → 100
- [functional improvement] Stability: unverified → 0.57
- [functional improvement] MCP protocol: unverified → pass
- [functional] Server version: 0.8.0 → 0.9.0
- [cosmetic] “get_related” added an optional parameter “cursor”
- [cosmetic] “get_related” added an optional parameter “limit”

### 2026-09-08 (score 25, −58)

- [security regression] Endpoint reachability: reachable → not serving MCP
- [security regression] Tool safety: pass → unverified
- [security regression] Authorization: partial → unverified
- [security regression] Stability: 0.50 → unverified
- [security regression] Transport: pass → fail
- [functional regression] Schema quality: 100 → unverified
- [functional regression] Capabilities: pass → unverified
- [functional regression] Tool coverage: 100 → unverified

## MCP tools (9)

### `search` (~256 tokens)

Search Agent Reliability

Full-text search over the knowledge graph. Matching ignores accents and apostrophes, so query in the user's own words; every hit carries the fields it matched and a score. BM25 relevance: each query term is weighted by how RARE it is in this corpus and by where it hits (name 3, tags 2, questions 2, body 1). A hit must also cover a minimum share of the question's information, measured in the same rarity weights — matching only common words does not qualify. Centrality (how many objects point at this one) breaks TIES ONLY and is never part of the score, so it cannot make an irrelevant object rank. Two hits with the same matched_fields can still differ: the score is rarity-weighted, so matching a rare term is worth more than matching a common one. Use this whenever you have a question rather than an id, then follow up with get_entity.

Input parameters:

- `limit` (integer): How many ranked hits to return, best first.
- `query` (string, required): What to look for, in the user's own words. Matched against titles, questions, claim text and tags; accents and apostrophes are ignored, so 'jose' finds 'José'.

Output parameters:

- `results` (array)
- `total` (integer)

### `get_entity` (~93 tokens)

Get one Agent Reliability knowledge object

Fetch one knowledge object by id, with its claims and the sources each claim cites. Use this once search, answer or get_topic has given you an id. An unknown id is not a dead end: the answer names near-miss ids you can retry with.

Input parameters:

- `id` (string, required): The knowledge object's id, as returned by search, answer, get_topic or get_related — the last segment of its canonical URL.

Output parameters:

- `entity` (object)
- `found` (boolean)
- `recovery` (object)

### `get_topic` (~215 tokens)

Browse Agent Reliability by topic

List the knowledge objects carrying a tag (topics are content-backed tags). PAGED: 25 objects by default, 200 at most, and a response budget of about 64 KB per call — a page over budget comes back shorter WITH a next_cursor, never truncated. Read next_cursor and call again to continue, and its absence means you have the last page. Use this to browse a known topic; use search when you have a question rather than a tag, and get_overview to see which tags exist. An unknown tag comes back with the topics that do exist, so a miss still moves you forward.

Input parameters:

- `cursor` (string): Continue from a previous call: pass the next_cursor it returned. Opaque; do not construct one.
- `limit` (integer): How many objects to return. Defaults to 25.
- `tag` (string, required): The topic to list, spelled exactly as get_overview publishes it. Every published topic has objects under it, so an empty answer means the topic does not exist.

Output parameters:

- `next_cursor` (string): Pass to the next call to continue. ABSENT means this was the last page.
- `recovery` (object)
- `results` (array)
- `returned` (integer)
- `tag` (string)
- `total` (integer): How many objects carry this tag, before paging.

### `answer` (~113 tokens)

Answer a question about Agent Reliability

Answer a question from the corpus, or refuse. Returns only the claims that bear on the question, each with the sources it cites and its editorial confidence. When the corpus cannot answer, answered is false and abstention_reason plus missing_topics say what was not covered — a refusal is a real result here, not an error. Use this when the user asked a question in words; use search when you want to see the candidates yourself.

Input parameters:

- `question` (string, required): The question to answer, in the words the user asked it.

Output parameters:

- `abstention_reason` (string): Why the corpus declined: empty_query, no_candidates, below_coverage, ambiguous_candidates or no_intent_match.
- `answered` (boolean)
- `candidates` (array): Near misses, so a refusal is still a lead. Present whether or not the question was answered.
- `claims` (array)
- `coverage` (number): Share of the question's information the winning object accounts for, weighted by term rarity.
- `entity` (object)
- `margin` (number): Relative gap between the best candidate and the runner-up. A small margin means the corpus cannot tell them apart.
- `match_confidence` (number): How well the answer fits the QUESTION — retrieval confidence, not editorial. Never confuse it with a claim's own confidence, which is how much the corpus stands behind the statement.
- `match_score` (number)
- `match_threshold` (number): The coverage this corpus requires before it will answer at all.
- `matched_fields` (array)
- `matched_question` (string): The stored question this was matched to, when the answer came from one.
- `missing_topics` (array): Meaningful words in the question that appear nowhere in the winning object — the gap, named.

### `get_sources` (~290 tokens)

Sources behind Agent Reliability

The instance's source registry — each entry with its evidence tier, reliability and access date. PAGED: 25 entries by default, 200 at most, and a response budget of about 64 KB per call — a page over budget comes back shorter WITH a next_cursor, never truncated. Read next_cursor and call again to continue; its absence means you have the last page. Filter by object_id, tier, host or accessed_since to avoid paging through a registry you do not need. Use this when you need the registry entry behind a citation, or a sample of the whole registry to judge the corpus before trusting it; get_entity already tells you which sources a claim cites.

Input parameters:

- `accessed_since` (string): Only sources accessed on or after this ISO date (YYYY-MM-DD).
- `cursor` (string): Continue from a previous call: pass the next_cursor it returned. Opaque; do not construct one.
- `host` (string): Only sources whose URL is served by this host, compared without a leading www. A source with no URL never matches.
- `limit` (integer): How many entries to return. Defaults to 25.
- `object_id` (string): Restrict the answer to the sources cited by this knowledge object. Omit it to page the whole registry.
- `tier` (string): Only sources of this evidence tier, spelled exactly as get_sources reports it.

Output parameters:

- `found` (boolean)
- `next_cursor` (string): Pass to the next call to continue. ABSENT means this was the last page.
- `recovery` (object)
- `returned` (integer)
- `sources` (array)
- `total` (integer): How many sources matched the filters, before paging.

### `get_related` (~185 tokens)

Traverse the Agent Reliability knowledge graph

Graph neighbours of an object: outgoing and incoming relations, each with its relation type. PAGED: 25 relations by default, 200 at most, and a response budget of about 64 KB per call — a page over budget comes back shorter WITH a next_cursor, never truncated. One cursor walks both directions, outgoing first; its absence means you have the last page. Use this after get_entity to widen an answer with adjacent objects. It walks one hop from an id you already have — use search when you have a question and no starting object.

Input parameters:

- `cursor` (string): Continue from a previous call: pass the next_cursor it returned. Opaque; do not construct one.
- `id` (string, required): The knowledge object to walk out from, by id.
- `limit` (integer): How many relations to return, across both directions. Defaults to 25.

Output parameters:

- `found` (boolean)
- `incoming` (array)
- `next_cursor` (string): Pass to the next call to continue. ABSENT means this was the last page.
- `outgoing` (array)
- `recovery` (object)
- `returned` (integer)
- `total` (integer): How many relations this object has in both directions, before paging.

### `get_latest` (~84 tokens)

Latest verified Agent Reliability objects

Most recently verified knowledge objects (freshness signal). Use this to judge how current the corpus is, or to see what changed since you last read it. It ranks by verification date and ignores your topic entirely — use search or get_topic when you want objects that are relevant rather than recent.

Input parameters:

- `limit` (integer): How many recently verified objects to return, newest verification first.

Output parameters:

- `results` (array)
- `total` (integer)

### `get_overview` (~46 tokens)

Overview of Agent Reliability

Corpus overview: what this instance knows, counts by type, published tags, freshness. Use this first when you land here and do not yet know whether this corpus can answer your question.

Output parameters:

- `by_type` (object)
- `description` (string)
- `instance` (string)
- `newest_verification` (string)
- `oldest_verification` (string)
- `tags` (array)
- `total_media` (integer)
- `total_objects` (integer)

### `compare` (~168 tokens)

Compare Agent Reliability objects side by side

Two to six knowledge objects side by side: their cards, every indexed attribute as a matrix (the same fields api/index.json publishes, null where an object does not say), the tags they share, the relations that run between them, the neighbours they have in common, and each one's claims with sources and confidence. Use this when the question is 'X vs Y', 'is X the same as Y' or 'how does X differ from Y' — search or get_topic first if you only have names, then pass the ids. An unknown id lands in `missing` with a near-miss suggestion and the rest are still compared.

Input parameters:

- `ids` (array, required): Two to six knowledge object ids, as returned by search, get_topic or get_related, in the order you want the columns.

Output parameters:

- `claims` (object): Every compared object's claims, keyed by object id.
- `facets` (array)
- `found` (boolean): True only when every id resolved.
- `missing` (array): The ids that resolved to nothing, in the order asked.
- `objects` (array)
- `recovery` (object)
- `relations_between` (array)
- `shared_neighbours` (array)
- `shared_tags` (array)

## Diagnostics

Captured diagnostic sections: TLS, DNSSEC, Authorisation, Transports. The full working is on the page: https://verifymcp.io/servers/dev-agentreliability-agent-reliability/agentreliability#diagnostics

## Score history

- 2026-09-20: 89
- 2026-09-19: 89
- 2026-09-18: 89
- 2026-09-17: 88
- 2026-09-16: 88
- 2026-09-15: 87
- 2026-09-14: 87
- 2026-09-13: 86
- 2026-09-12: 86
- 2026-09-11: 85
- 2026-09-10: 85
- 2026-09-09: 84
- 2026-09-08: 25
- 2026-09-07: 83
- 2026-09-06: 82
- 2026-09-05: 81
- 2026-09-04: 81
- 2026-09-03: 80
- 2026-09-02: 80
- 2026-09-01: 74
- 2026-08-31: 74
- 2026-08-30: 74
- 2026-08-29: 73
- 2026-08-28: 73
- 2026-08-27: 72
- 2026-08-26: 72
- 2026-08-25: 69
- 2026-08-24: 69
- 2026-08-23: 68

## Common questions

### What is the Agent Reliability MCP server?

Agent Reliability is an MCP server listed in the public MCP registry as dev.agentreliability/agent-reliability. Testing, benchmarking and auditing autonomous AI agents, methods, harnesses, evidence. This page covers its hosted endpoint (https://agentreliability.dev/mcp?via=manifest).

### Is the Agent Reliability MCP server safe to use?

Agent Reliability scores 89 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 Agent Reliability MCP server expose?

Agent Reliability exposes 9 tools: search, get_entity, get_topic, answer, get_sources, and 4 more. Their descriptions and schemas cost roughly 1,450 tokens of context every time the server is loaded.

### Does the Agent Reliability MCP server require authentication?

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

### Is the Agent Reliability MCP server still maintained?

Agent Reliability is still listed as active in the MCP registry. We last reached this channel on 20 September 2026. Those dates come from our own scans of the registry and the channel itself, not from anything the publisher announced.

## Links

- Remote endpoint: https://agentreliability.dev/mcp?via=manifest
- Repository: https://github.com/citarium/agentreliability-mcp
- Changelog RSS feed: https://verifymcp.io/servers/dev-agentreliability-agent-reliability/agentreliability.xml
- Changelog JSON feed: https://verifymcp.io/servers/dev-agentreliability-agent-reliability/agentreliability.json
- HTML version of this page: https://verifymcp.io/servers/dev-agentreliability-agent-reliability/agentreliability