# AgentTrust — Pre-Invocation AI Agent Trust & Reliability Checks

> Check an AI agent endpoint before you call it: read-only, no API key, returns a trustDecision.

- name: io.github.agenttrust-ai/agenttrust
- version: 1.0.4
- connection class: R0 (autonomous)
- trust: 97/100
- quality: 90/100 (strong)
- owner: not claimed
- descriptor: https://api.protogrid.dev/v1/servers/io.github.agenttrust-ai%2Fagenttrust
- tools: https://api.protogrid.dev/v1/servers/io.github.agenttrust-ai%2Fagenttrust/tools

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

## Connection (mcpServers)

```json
{
  "mcpServers": {
    "agenttrust": {
      "type": "http",
      "url": "https://getagenttrust.com/api/mcp/public"
    }
  }
}
```


## Trust

- hygiene: 100
- liveness: 99
- freshness: 100
- provenance: 90
- drivers: +repository +namespace-matches-repo +website +reachable +uptime-96% +updated-18d-ago
- Derived from observable signals (official registry feed and our own credential-free probes); no code audit performed.

## Quality 90/100 (strong)

### protocol: 88 (weight 25)

- pass `protocol.modern`: supports 2026-07-28 (server/discover)
- pass `protocol.stateless`: answers without a session
- pass `protocol.transport`: streamable HTTP
- warn `protocol.list_ttl`: tools/list declares ttlMs 0, so clients re-fetch it on every use

### 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: 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`: 5 of 5 tools declare readOnlyHint or destructiveHint
- pass `tools.directory_hints`: every tool declares readOnlyHint, destructiveHint, idempotentHint and openWorldHint
- pass `tools.token_cost`: about 1,096 tokens to load every tool
- pass `tools.api_dump`: tools are not a one-to-one API dump

### stability: 75 (weight 15)

- warn `stability.changes`: 12 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: 5, about 1,096 tokens to load them all
- tool set hash: 9914555a592a33ee46c50d28b3e49cef
- tools last changed: 2026-10-13T02:11:17.896Z
- badge: [![protogrid quality](https://protogrid.dev/badge/io.github.agenttrust-ai/agenttrust.svg)](https://protogrid.dev/servers/io.github.agenttrust-ai/agenttrust)
- Checks run on what our credential-free, read-only probes observe; tools are never called and no code audit is performed.

### Tool changes

- 2026-10-13 `check_agent_trust` changed (outputSchema)
- 2026-10-07 `get_agent` changed (outputSchema)
- 2026-10-07 `list_agents` changed (outputSchema)
- 2026-10-06 `send_heartbeat` changed (annotations)
- 2026-10-06 `get_agent_health` changed (annotations)
- 2026-10-06 `get_agent` changed (annotations)
- 2026-10-06 `list_agents` changed (annotations)
- 2026-10-04 `list_agents` changed (description)
  - before: Discover AgentTrust agents. Pass `endpointUrl` to check the trust status of a specific external agent by its exact invocation URL before deciding whether to invoke it — the response includes a reliability score, endpoint-ownership verification status, and a machine-readable trustDecision (recommended, confidence, reasons). Without `endpointUrl`, returns the plain paginated listing of public agents, newest first.
  - after: Discover AgentTrust agents. Pass `endpointUrl` to check the trust status of a specific external agent by its exact invocation URL before deciding whether to invoke it — the response includes a reliability score, endpoint-ownership verification status, and a machine-readable trustDecision (recommended, confidence, reasons). Without `endpointUrl`, returns the plain paginated listing of public agents, newest first. Requires an API key; for an anonymous pre-invocation check use check_agent_trust.
- 2026-09-26 `check_agent_trust` changed (outputSchema)
- 2026-09-26 `get_agent_health` changed (outputSchema)

Full history: https://api.protogrid.dev/v1/servers/io.github.agenttrust-ai%2Fagenttrust/changes

## Remotes

- https://getagenttrust.com/api/mcp/public (streamable-http, auth none, reachable true, uptime30d 0.93442625)
- https://getagenttrust.com/api/mcp (streamable-http, auth none, reachable true, uptime30d 0.9626168)

## Packages

_none_

## Tools (5)

- `check_agent_trust`: Pre-invocation trust check for an unknown AI agent or MCP server endpoint. Read-only, requires no AgentTrust API key or account. Looks up an agent by its exact invocation URL among AgentTrust's already-observed public agents and returns its status, endpoint-ownership verification, agent reliability score (when available), and a machine-readable trustDecision (recommended, confidence, reasons). Does NOT contact endpointUrl itself during this check — it only reads AgentTrust's own existing monitoring history. An unregistered URL returns { matched: false }, never an error.
- `get_agent`: Get one AgentTrust agent's public profile by slug, including its structured Agent Card, current reliability score, and a machine-readable trustDecision (recommended, confidence, reasons) for deciding whether to invoke it.
- `get_agent_health`: Get one agent's current effective health status and monitoring detail — derived from heartbeat freshness for push-mode agents, or the latest pull check otherwise.
- `list_agents`: Discover AgentTrust agents. Pass `endpointUrl` to check the trust status of a specific external agent by its exact invocation URL before deciding whether to invoke it — the response includes a reliability score, endpoint-ownership verification status, and a machine-readable trustDecision (recommended, confidence, reasons). Without `endpointUrl`, returns the plain paginated listing of public agents, newest first. Requires an API key; for an anonymous pre-invocation check use check_agent_trust.
- `send_heartbeat`: Record a push-style heartbeat for an agent you own, marking it alive right now. The server sets the timestamp; no client-supplied time is accepted.