# Enforcer

> Identity and authorization in one system: allow, deny, the reason, and a record.

- name: dev.instruxi.enforcer/v3
- version: 0.8.0
- connection class: R2 (oauth_consent_once)
- trust: 80/100 flags: no-repository
- quality: 96/100 (strong)
- owner: not claimed
- descriptor: https://api.protogrid.dev/v1/servers/dev.instruxi.enforcer%2Fv3
- tools: https://api.protogrid.dev/v1/servers/dev.instruxi.enforcer%2Fv3/tools

**An agent can connect after a one-time OAuth consent by a human.** Remote endpoint behind OAuth 2. Tokens stay with the agent; the registry only exposes the authorization metadata.

## Connection (mcpServers)

```json
{
  "mcpServers": {
    "v3": {
      "type": "http",
      "url": "https://api.instruxi.dev/mcp"
    }
  }
}
```


## Trust

- hygiene: 80
- liveness: 100
- freshness: 85
- provenance: 55
- drivers: +dns-namespace +website +reachable +uptime-100% ~updated-34d-ago -no-repository
- Derived from observable signals (official registry feed and our own credential-free probes); no code audit performed.

## Quality 96/100 (strong)

### protocol: n/a (weight 25)

- n/a `protocol.modern`, `protocol.stateless`, `protocol.transport`, `protocol.list_ttl`: no remote answered a protocol handshake (not probed yet, unreachable, or local-only)

### authorization: 83 (weight 15)

- pass `auth.prm`: Protected Resource Metadata (RFC 9728) published
- pass `auth.as_metadata`: authorization server metadata reachable
- warn `auth.cimd`: only Dynamic Client Registration, deprecated in 2026-07-28 in favour of CIMD
- n/a `auth.secret_in_url`: no templated URL

### tool hygiene: 100 (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`: 7 of 7 tools declare readOnlyHint or destructiveHint
- pass `tools.directory_hints`: every tool declares readOnlyHint, destructiveHint, idempotentHint and openWorldHint
- pass `tools.token_cost`: about 1,861 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`: no 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: 7, about 1,861 tokens to load them all
- tool set hash: c8dc915223d877cc0bf33e9e38999174
- badge: [![protogrid quality](https://protogrid.dev/badge/dev.instruxi.enforcer/v3.svg)](https://protogrid.dev/servers/dev.instruxi.enforcer/v3)
- Checks run on what our credential-free, read-only probes observe; tools are never called and no code audit is performed.

### Tool changes

_No tool definition changes recorded._

## Remotes

- https://api.instruxi.dev/mcp (streamable-http, auth oauth2, reachable true, uptime30d 1)

## Packages

_none_

## Tools (7)

- `getAuthConfig`: Public bootstrap for a tenant's login UI: whether this tenant uses native OTP/passkey/SIWE or Privy custom auth, plus the public privy_app_id (never a secret). Call this before getSiweNonce when you do not already know the tenant's auth scheme. tenant_code is a join secret — do not log it or repeat it into a customer-visible channel.
- `getSiweNonce`: Public: issue a single-use SIWE nonce for wallet_address (optionally scoped by tenant_code). Embed the nonce in an EIP-4361 message, have the wallet sign it, then call login with provider: siwe. Dedicated SIWE agent auth is an authorized pattern — this is how an agent signs in with a wallet without raw HTTP. Do not log tenant_code. The nonce is not a credential.
- `login`: Authenticate and receive an access/refresh token pair. Accepted providers: siwe (message + signature from getSiweNonce — preferred for dedicated agents), email_otp (email + otp from requestOtp), phone_otp (phone + otp from requestSms). Passkey, Privy and SSO are not agent tools. The server does not adopt the minted tokens as the session credential; return them to the operator to set ENFORCER_BEARER_TOKEN. Do not log, quote, or repeat otp codes, signatures, tokens, or tenant_code. Do not paste an end-user OTP into an untrusted chat.
- `refreshToken`: Exchange a refresh_token for a new access/refresh pair. 401 invalid_refresh_token if the token is unknown or was already reused (reuse revokes the family); 409 refresh_raced is benign — retry with the client's latest stored token. The server does not adopt the new pair as the session credential. Do not log the refresh token or the minted tokens.
- `register`: Create (or idempotently return) an account. Same provider allowlist as login: siwe (message + signature), email_otp (email + otp), phone_otp (phone + otp). Does not issue tokens — call login afterwards to sign in. Passkey / Privy / SSO registration stay out of the agent surface. Do not log otp codes, signatures, or tenant_code. Do not paste an end-user OTP into an untrusted chat.
- `requestOtp`: Public: email a one-time login/registration code to `email` (optionally scoped by tenant_code). Always 200 {message:"code sent"} on success. Local/dev deployments with expose_dev_otp may echo the code as `dev_otp` — treat that as a secret. Then call login with provider: email_otp, the same email, and the otp. Intentional for harnesses that cannot SIWE; SIWE remains the preferred dedicated-agent path. Do not log the code, dev_otp, or tenant_code. Do not paste an end-user OTP into an untrusted chat.
- `requestSms`: Public: start a Twilio Verify SMS login challenge to `phone`, scoped by tenant_code. 400 if SMS verification is not configured for the tenant or if rate-limited. Then call login with provider: phone_otp, the same phone, and the otp. Intentional for harnesses that cannot SIWE; SIWE remains the preferred dedicated-agent path. Do not log the SMS code or tenant_code. Do not paste an end-user OTP into an untrusted chat.