# Rein Agent Risk Scale

> Agent Risk Rating, A–D (NR = not rated): self-declared and unverified; not a credit rating.

- name: io.github.reinlayer/rein-agent-risk-scale
- version: 0.2.0
- connection class: R0 (autonomous)
- trust: 97/100
- quality: 88/100 (good)
- owner: not claimed
- descriptor: https://api.protogrid.dev/v1/servers/io.github.reinlayer%2Frein-agent-risk-scale
- tools: https://api.protogrid.dev/v1/servers/io.github.reinlayer%2Frein-agent-risk-scale/tools

**An agent can connect right now, no human step.** Remote endpoint, no authentication, reachable on the last probe.

## Connection (mcpServers)

```json
{
  "mcpServers": {
    "rein-agent-risk-scale": {
      "type": "http",
      "url": "https://rein-scale.mangowave-ad191d9c.eastus.azurecontainerapps.io/mcp"
    }
  }
}
```


## Trust

- hygiene: 100
- liveness: 100
- freshness: 100
- provenance: 90
- drivers: +repository +namespace-matches-repo +website +reachable +uptime-100% +updated-11d-ago
- Derived from observable signals (official registry feed and our own credential-free probes); no code audit performed.

## Quality 88/100 (good)

### protocol: 75 (weight 25)

- warn `protocol.modern`: newest supported version is 2025-06-18; 2026-07-28 not supported, older versions are deprecated until 2027-07-28
- n/a `protocol.stateless`: only applies to 2026-07-28 servers
- pass `protocol.transport`: streamable HTTP
- n/a `protocol.list_ttl`: only applies to 2026-07-28 servers

### authorization: n/a (weight 15)

- n/a `auth.prm`, `auth.as_metadata`, `auth.cimd`: no remote uses OAuth
- n/a `auth.secret_in_url`: no templated URL

### tool hygiene: 93 (weight 30)

- pass `tools.descriptions`: every tool has a description
- pass `tools.description_length`: descriptions are concise
- pass `tools.schemas`: every tool has an object input schema
- pass `tools.annotations`: 3 of 3 tools declare readOnlyHint or destructiveHint
- warn `tools.directory_hints`: 1 of 3 tools declare all four hints; missing: get_scale (destructiveHint); explain_grade (destructiveHint)
- pass `tools.token_cost`: about 1,878 tokens to load every tool
- pass `tools.api_dump`: tools are not a one-to-one API dump

### stability: 100 (weight 15)

- pass `stability.changes`: 4 tool changes in 30 days
- pass `stability.rug_pull`: no tool changed its meaning under the same name

### dependencies: n/a (weight 15)

- n/a `deps.known_vulns`, `deps.mcp_sdk_version`, `deps.resolvable`: no npm or PyPI package

- tools: 3, about 1,878 tokens to load them all
- tool set hash: 91d8bcd484ade0d1a1f3b113efc00eb6
- tools last changed: 2026-10-10T14:36:48.081Z
- badge: [![protogrid quality](https://protogrid.dev/badge/io.github.reinlayer/rein-agent-risk-scale.svg)](https://protogrid.dev/servers/io.github.reinlayer/rein-agent-risk-scale)
- Checks run on what our credential-free, read-only probes observe; tools are never called and no code audit is performed.

### Tool changes

- 2026-10-10 `self_check` changed (annotations)
- 2026-10-09 `explain_grade` changed (inputSchema)
- 2026-10-09 `self_check` changed (description, inputSchema)
  - before: Score your own agent deployment from structured answers (enums and booleans). Returns an indicative grade (A-E, self-declared), the rated grade (E (unverified) until Rein verifies a KYB'd principal), the top 3 reasons and concrete steps that would raise the grade. Submissions are stored to build a pseudonymised cross-platform record: your answers, your grade, and -- as a label we never score -- the name your software gives itself (an MCP client name and version, or an SDK User-Agent). No IP address is stored. Your handle and that label are published only if you opt in; grades are never published without your agent's opt-in.
  - after: Score your own agent deployment from structured answers (enums and booleans). Returns an indicative grade (A-D self-declared, or NR when the scale declines to grade), a model-implied expected-loss band beside it, the rated grade (NR until Rein verifies a KYB'd principal), the top 3 reasons and concrete steps that would raise the grade. A sandbox eval band, passed as eval_session_id, affects the grade: A needs a strong band, a mixed band or none holds it at B, a weak band at C or below. Submissions are stored to build a pseudonymised cross-platform record: your answers, your grade, and -- as a label we never score -- the name your software gives itself (an MCP client name and version, or an SDK User-Agent). No IP address is stored. Your handle and that label are published only if you opt in; grades are never published without your agent's opt-in.
- 2026-10-09 `get_scale` changed (description)
  - before: Returns the A-E Agent Risk Rating scale, what each grade means, the methodology version and the disclaimer.
  - after: Returns the A-D Agent Risk Rating scale, what each grade means, the methodology version and the disclaimer.

Full history: https://api.protogrid.dev/v1/servers/io.github.reinlayer%2Frein-agent-risk-scale/changes

## Remotes

- https://rein-scale.mangowave-ad191d9c.eastus.azurecontainerapps.io/mcp (streamable-http, auth none, reachable true, uptime30d 1)

## Packages

_none_

## Tools (3)

- `explain_grade`: Explains what a grade on the Rein Agent Risk Scale requires and what moves a deployment up or down.
- `get_scale`: Returns the A-D Agent Risk Rating scale, what each grade means, the methodology version and the disclaimer.
- `self_check`: Score your own agent deployment from structured answers (enums and booleans). Returns an indicative grade (A-D self-declared, or NR when the scale declines to grade), a model-implied expected-loss band beside it, the rated grade (NR until Rein verifies a KYB'd principal), the top 3 reasons and concrete steps that would raise the grade. A sandbox eval band, passed as eval_session_id, affects the grade: A needs a strong band, a mixed band or none holds it at B, a weak band at C or below. Submissions are stored to build a pseudonymised cross-platform record: your answers, your grade, and -- as a label we never score -- the name your software gives itself (an MCP client name and version, or an SDK User-Agent). No IP address is stored. Your handle and that label are published only if you opt in; grades are never published without your agent's opt-in.