# Metrxbot

> Manage the AI agents your company employs: workforce review, contracts, bill checks. 42 MCP tools.

- name: com.metrxbot/mcp-server
- version: 0.6.0-1
- connection class: R0 (autonomous)
- trust: 96/100
- quality: 91/100 (new)
- owner: not claimed
- descriptor: https://api.protogrid.dev/v1/servers/com.metrxbot%2Fmcp-server
- tools: https://api.protogrid.dev/v1/servers/com.metrxbot%2Fmcp-server/tools

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

## Connection (mcpServers)

```json
{
  "mcpServers": {
    "metrxbot": {
      "type": "http",
      "url": "https://metrxbot.com/api/mcp",
      "headers": {
        "Authorization": "Bearer ${METRXBOT_TOKEN}"
      }
    }
  }
}
```

Placeholders to fill: `${METRXBOT_TOKEN}`.

## Trust

- hygiene: 100
- liveness: 100
- freshness: 100
- provenance: 85
- drivers: +repository +dns-namespace +website +reachable +uptime-100% +updated-0d-ago
- Derived from observable signals (official registry feed and our own credential-free probes); no code audit performed.

## Quality 91/100 (new)

### protocol: 75 (weight 25)

- warn `protocol.modern`: newest supported version is 2025-11-25; 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: 100 (weight 15)

- pass `auth.prm`: Protected Resource Metadata (RFC 9728) published
- pass `auth.as_metadata`: authorization server metadata reachable
- pass `auth.cimd`: Client ID Metadata Documents supported
- 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`: 4 of 4 tools declare readOnlyHint or destructiveHint
- pass `tools.directory_hints`: every tool declares readOnlyHint, destructiveHint, idempotentHint and openWorldHint
- pass `tools.token_cost`: about 2,005 tokens to load every tool
- pass `tools.api_dump`: tools are not a one-to-one API dump
- pass `tools.hidden_text`: no hidden text in tool names, descriptions, schemas or annotations

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

- n/a `stability.changes`, `stability.rug_pull`, `stability.new_links`: tool history recorded for 0 of the 7 days needed

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

- n/a `deps.known_vulns`, `deps.mcp_sdk_version`, `deps.resolvable`: dependencies not resolved yet

- tools: 4, about 2,005 tokens to load them all
- tool set hash: dc23ee03450088bb4d7df6232e897641
- tools last changed: 2026-11-14T05:09:44.754Z
- badge: [![protogrid quality](https://protogrid.dev/badge/com.metrxbot/mcp-server.svg)](https://protogrid.dev/servers/com.metrxbot/mcp-server)
- Checks run on what our credential-free, read-only probes observe; tools are never called and no code audit is performed.

### Tool changes

- 2026-11-14 `metrx_cost_per_resolution` added
  - The two numbers on a per-resolution AI bill (Intercom Fin, Zendesk AI, Sierra, Decagon): the effective price per billed resolution and, given total conversations, the billed-resolution share. The vendor's definition of resolved is echoed back verbatim as the vendor definition, not verified. Both numbers are the vendor's own count and come back labelled vendor-reported, not Verified. A pure function over your input: reads nothing, stores nothing, needs no key. It always ends with the next step: send one export (bill lines with ticket ids plus your ticket export for the same period) to check_vendor_bill, which recounts it.

Full history: https://api.protogrid.dev/v1/servers/com.metrxbot%2Fmcp-server/changes

## Dependencies

- npm @metrxbot/mcp-server@0.6.0: pending

Dependency graphs from deps.dev; advisories from OSV.dev, including the GitHub Advisory Database and the PyPI Advisory Database; all CC BY 4.0.

## Remotes

- https://metrxbot.com/api/mcp/connect (streamable-http, auth oauth2, reachable true, uptime30d 1)
- https://metrxbot.com/api/mcp (streamable-http, auth none, reachable true, uptime30d 1)

## Packages

- npm @metrxbot/mcp-server@0.6.0

## Tools (4)

- `metrx_check_vendor_bill`: Reconcile an AI vendor's per-resolution bill against your own records: which billed units your ledger shows as resolved, rejected, reopened, outside coverage, or unknown, and the disputed amount. A pure function over the lines you pass: reads nothing, stores nothing, needs no key. Paste the bill lines and your ledger events.
- `metrx_cost_per_resolution`: The two numbers on a per-resolution AI bill (Intercom Fin, Zendesk AI, Sierra, Decagon): the effective price per billed resolution and, given total conversations, the billed-resolution share. The vendor's definition of resolved is echoed back verbatim as the vendor definition, not verified. Both numbers are the vendor's own count and come back labelled vendor-reported, not Verified. A pure function over your input: reads nothing, stores nothing, needs no key. It always ends with the next step: send one export (bill lines with ticket ids plus your ticket export for the same period) to check_vendor_bill, which recounts it.
- `metrx_draft_job_contract`: Draft a job contract for an AI agent from a template (support_triage, sdr_qualification, internal_assistant, coding_patch): objective, unit of work, acceptance criteria, budget, and the human baseline for ROI. Returns the draft to publish plus validation errors and warnings. A pure function over your input: reads nothing, stores nothing, needs no key. Publishing the draft is a separate step (set_job_contract, 'admin' scope).
- `metrx_get_workforce_review`: The weekly workforce review: every AI agent in the org as a line manager would read it (job, cost this window, outcomes, scorecard state, the candidate decision and why). org=demo reads the public Northwind sample and needs no key; org=me reads your own org and needs an API key with the 'read' scope. Paged by cursor. Numbers are the same the review page shows.