switchboard-100.pages.dev WebMCP Tools, Schemas & Agent Capabilities

switchboard-100.pages.dev

live Unverified (seed import) AI & Agents 15 tools imperative implementation

Human + agent control center for LLM provider budgets, priced model catalogs and routing policy. Agent-ready via WebMCP.

First seen 2026-11-05 · Last seen 2026-11-05 · Source: webmcp.com

Answer 11 · Act 3 · Transact 1 · Score: Not yet scored

Tool inspector

list_models

answer imperative

List the priced model catalogue joined with this operator's own measurements. Prices are per million tokens from published provider documentation. Call get_provenance for how the measurements were produced and how coarse they are. Per model: requestSuccessRate (0-1), medianLatencyMs, p95LatencyMs, observedCalls (the sample size behind those latencies — check it before trusting a small latency gap), costPer1kSuccessfulUsd (cost per 1000 DELIVERED outputs, i.e. price divided by success rate; prefer it over raw price, because a cheap model that fails half its calls is not cheap), and meetsQualityGates. GOVERNANCE, and read this carefully: on a MODEL, dataRetentionDays and trainsOnData are null when the provider does not document them, and null means UNKNOWN, which is not the same as safe. On a TRAFFIC CLASS the same-named constraints being null means UNCONSTRAINED. The two nulls are opposites, so do not carry one across the join. Every governance filter here excludes undocumented models rather than assuming the best. If you only need a set of models that is legal for a class, pass eligibleForClassId and skip the individual filters — it applies the class's real constraints, including the ones these filters cannot express.

Input schema

Property	Type	Required	Description
`providerId`	string	no	Restrict to one provider, e.g. 'opencode-go'.
`measuredOnly`	boolean	no	Only models with measurements. Default false.
`excludeDead`	boolean	no	Drop models that returned zero successful calls in every run. Unmeasured models are kept — use measuredOnly to drop those. Default false.
`maxDataRetentionDays`	number	no	Only models documented to retain data for at most this many days. Models with undocumented retention are excluded.
`excludeTrainsOnData`	boolean	no	Drop models that train on submitted data AND models where this is undocumented, since undocumented is not safe. Default false.
`minSuccessRate`	number	no	Only models with a measured request success rate at or above this (0-1). Excludes unmeasured models.
`maxCostPer1kDeliveredUsd`	number	no	Only models at or below this cost per 1000 delivered outputs, in USD. Excludes unmeasured models.
`limit`	number	no	Cap the number of models returned, cheapest per delivered output first. Omit for all matches.
`maxP95Ms`	number	no	Only models with measured p95 latency at or below this many milliseconds.
`requireQualityGates`	boolean	no	Only models that passed the quality gates. Default false.
`eligibleForClassId`	string	no	Return only models satisfying every constraint of this traffic class. Simplest way to get a legal candidate set.

Raw JSON schema

{
  "type": "object",
  "properties": {
    "providerId": {
      "type": "string",
      "description": "Restrict to one provider, e.g. 'opencode-go'."
    },
    "measuredOnly": {
      "type": "boolean",
      "description": "Only models with measurements. Default false."
    },
    "excludeDead": {
      "type": "boolean",
      "description": "Drop models that returned zero successful calls in every run. Unmeasured models are kept — use measuredOnly to drop those. Default false."
    },
    "maxDataRetentionDays": {
      "type": "number",
      "description": "Only models documented to retain data for at most this many days. Models with undocumented retention are excluded."
    },
    "excludeTrainsOnData": {
      "type": "boolean",
      "description": "Drop models that train on submitted data AND models where this is undocumented, since undocumented is not safe. Default false."
    },
    "minSuccessRate": {
      "type": "number",
      "description": "Only models with a measured request success rate at or above this (0-1). Excludes unmeasured models."
    },
    "maxCostPer1kDeliveredUsd": {
      "type": "number",
      "description": "Only models at or below this cost per 1000 delivered outputs, in USD. Excludes unmeasured models."
    },
    "limit": {
      "type": "number",
      "description": "Cap the number of models returned, cheapest per delivered output first. Omit for all matches."
    },
    "maxP95Ms": {
      "type": "number",
      "description": "Only models with measured p95 latency at or below this many milliseconds."
    },
    "requireQualityGates": {
      "type": "boolean",
      "description": "Only models that passed the quality gates. Default false."
    },
    "eligibleForClassId": {
      "type": "string",
      "description": "Return only models satisfying every constraint of this traffic class. Simplest way to get a legal candidate set."
    }
  }
}