noemaai.com WebMCP Tools, Schemas & Agent Capabilities

noemaai.com

live Unverified (seed import) AI & Agents 12 tools imperative implementation

Set up a Noema project on the web. Choose its purpose, device and local model, then scan the QR code and the project opens in Noema on your phone.

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

Answer 2 · Act 10 · Transact 0 · Score: Not yet scored

Tool inspector

configure_model_settings

act imperative

Patch one or more visible settings for the selected model without changing unrelated preferences. Omitted settings remain untouched. Set seed or any runtime override to null to clear its Blueprint override. resetToDefaults resets this webpage draft.

Page: /setup

Input schema

Property	Type	Required	Description
`temperature`	number	no	0 is focused and deterministic; 2 is highly varied.
`topP`	number	no	—
`topK`	integer	no	—
`minP`	number	no	—
`repetitionPenalty`	number	no	—
`repeatLastN`	integer	no	—
`presencePenalty`	number	no	—
`frequencyPenalty`	number	no	—
`reasoningEnabled`	boolean	no	—
`promptCacheEnabled`	boolean	no	—
`seed`	—	no	Set an integer for repeatable sampling, or null for a random seed.
`gpuLayers`	—	no	GPU layers from -1 (all/automatic) through 4096; null clears the Blueprint override.
`cpuThreads`	—	no	CPU thread count; null clears the Blueprint override.
`evaluationBatchSize`	—	no	Evaluation batch size, a multiple of 32; null clears the Blueprint override.
`physicalBatchSize`	—	no	Physical batch size, a multiple of 32 and no larger than the resolved evaluation batch (512 when omitted); null clears the Blueprint override.
`loadVisionProjector`	—	no	Load the model's vision projector when present; null clears the Blueprint override.
`kvCacheOffload`	—	no	Offload the KV cache when supported; null clears the Blueprint override.
`unifiedKVCache`	—	no	Use a unified KV cache; null clears the Blueprint override.
`keepInMemory`	—	no	Keep model weights in memory; null clears the Blueprint override.
`useMmap`	—	no	Memory-map model weights; null clears the Blueprint override.
`disableWarmup`	—	no	Skip model warmup; null clears the Blueprint override.
`preserveThinking`	—	no	Preserve thinking output for supported models; null clears the Blueprint override.
`flashAttention`	—	no	Use Flash Attention when supported; V-cache quantization is relevant when enabled. Null clears the Blueprint override.
`kCacheQuant`	—	no	K-cache quantization; null clears the Blueprint override.
`vCacheQuant`	—	no	V-cache quantization, relevant with Flash Attention; null clears the Blueprint override.
`moeActiveExperts`	—	no	Active MoE experts; null clears the Blueprint override.
`overfitMode`	—	no	Overfit handling, limited to off or automatic; null clears the Blueprint override.
`speculativeMode`	—	no	Speculative decoding mode, limited to off or MTP; null clears the Blueprint override.
`mtpAutoTune`	—	no	Automatically tune MTP drafting; relevant when speculativeMode is mtp. Null clears the Blueprint override.
`mtpDraftNMax`	—	no	Maximum MTP draft tokens, relevant for MTP; null clears the Blueprint override.
`mtpDraftNMin`	—	no	Minimum MTP draft tokens, no larger than the resolved maximum (2 when omitted); null clears the Blueprint override.
`mtpDraftPMin`	—	no	Minimum MTP draft probability from 0 through 1; null clears the Blueprint override.
`resetToDefaults`	boolean	no	Reset the webpage draft's required sampling settings and clear seed plus all optional advanced Blueprint overrides before applying supplied fields.

Raw JSON schema

{
  "type": "object",
  "properties": {
    "temperature": {
      "type": "number",
      "minimum": 0,
      "maximum": 2,
      "description": "0 is focused and deterministic; 2 is highly varied."
    },
    "topP": {
      "type": "number",
      "minimum": 0,
      "maximum": 1
    },
    "topK": {
      "type": "integer",
      "minimum": 1,
      "maximum": 2048
    },
    "minP": {
      "type": "number",
      "minimum": 0,
      "maximum": 1
    },
    "repetitionPenalty": {
      "type": "number",
      "minimum": 0.8,
      "maximum": 2
    },
    "repeatLastN": {
      "type": "integer",
      "minimum": 0,
      "maximum": 4096
    },
    "presencePenalty": {
      "type": "number",
      "minimum": -2,
      "maximum": 2
    },
    "frequencyPenalty": {
      "type": "number",
      "minimum": -2,
      "maximum": 2
    },
    "reasoningEnabled": {
      "type": "boolean"
    },
    "promptCacheEnabled": {
      "type": "boolean"
    },
    "seed": {
      "anyOf": [
        {
          "type": "integer",
          "minimum": 0,
          "maximum": 2147483647
        },
        {
          "type": "null"
        }
      ],
      "description": "Set an integer for repeatable sampling, or null for a random seed."
    },
    "gpuLayers": {
      "anyOf": [
        {
          "type": "integer",
          "minimum": -1,
          "maximum": 4096
        },
        {
          "type": "null"
        }
      ],
      "description": "GPU layers from -1 (all/automatic) through 4096; null clears the Blueprint override."
    },
    "cpuThreads": {
      "anyOf": [
        {
          "type": "integer",
          "minimum": 1,
          "maximum": 256
        },
        {
          "type": "null"
        }
      ],
      "description": "CPU thread count; null clears the Blueprint override."
    },
    "evaluationBatchSize": {
      "anyOf": [
        {
          "type": "integer",
          "minimum": 32,
          "maximum": 8192,
          "multipleOf": 32
        },
        {
          "type": "null"
        }
      ],
      "description": "Evaluation batch size, a multiple of 32; null clears the Blueprint override."
    },
    "physicalBatchSize": {
      "anyOf": [
        {
          "type": "integer",
          "minimum": 32,
          "maximum": 8192,
          "multipleOf": 32
        },
        {
          "type": "null"
        }
      ],
      "description": "Physical batch size, a multiple of 32 and no larger than the resolved evaluation batch (512 when omitted); null clears the Blueprint override."
    },
    "loadVisionProjector": {
      "anyOf": [
        {
          "type": "boolean"
        },
        {
          "type": "null"
        }
      ],
      "description": "Load the model's vision projector when present; null clears the Blueprint override."
    },
    "kvCacheOffload": {
      "anyOf": [
        {
          "type": "boolean"
        },
        {
          "type": "null"
        }
      ],
      "description": "Offload the KV cache when supported; null clears the Blueprint override."
    },
    "unifiedKVCache": {
      "anyOf": [
        {
          "type": "boolean"
        },
        {
          "type": "null"
        }
      ],
      "description": "Use a unified KV cache; null clears the Blueprint override."
    },
    "keepInMemory": {
      "anyOf": [
        {
          "type": "boolean"
        },
        {
          "type": "null"
        }
      ],
      "description": "Keep model weights in memory; null clears the Blueprint override."
    },
    "useMmap": {
      "anyOf": [
        {
          "type": "boolean"
        },
        {
          "type": "null"
        }
      ],
      "description": "Memory-map model weights; null clears the Blueprint override."
    },
    "disableWarmup": {
      "anyOf": [
        {
          "type": "boolean"
        },
        {
          "type": "null"
        }
      ],
      "description": "Skip model warmup; null clears the Blueprint override."
    },
    "preserveThinking": {
      "anyOf": [
        {
          "type": "boolean"
        },
        {
          "type": "null"
        }
      ],
      "description": "Preserve thinking output for supported models; null clears the Blueprint override."
    },
    "flashAttention": {
      "anyOf": [
        {
          "type": "boolean"
        },
        {
          "type": "null"
        }
      ],
      "description": "Use Flash Attention when supported; V-cache quantization is relevant when enabled. Null clears the Blueprint override."
    },
    "kCacheQuant": {
      "anyOf": [
        {
          "type": "string",
          "enum": [
            "F32",
            "F16",
            "Q8_0",
            "Q5_0",
            "Q5_1",
            "Q4_0",
            "Q4_1"
          ]
        },
        {
          "type": "null"
        }
      ],
      "description": "K-cache quantization; null clears the Blueprint override."
    },
    "vCacheQuant": {
      "anyOf": [
        {
          "type": "string",
          "enum": [
            "F32",
            "F16",
            "Q8_0",
            "Q5_0",
            "Q5_1",
            "Q4_0",
            "Q4_1",
            "IQ4_NL"
          ]
        },
        {
          "type": "null"
        }
      ],
      "description": "V-cache quantization, relevant with Flash Attention; null clears the Blueprint override."
    },
    "moeActiveExperts": {
      "anyOf": [
        {
          "type": "integer",
          "minimum": 1,
          "maximum": 256
        },
        {
          "type": "null"
        }
      ],
      "description": "Active MoE experts; null clears the Blueprint override."
    },
    "overfitMode": {
      "anyOf": [
        {
          "type": "string",
          "enum": [
            "off",
            "automatic"
          ]
        },
        {
          "type": "null"
        }
      ],
      "description": "Overfit handling, limited to off or automatic; null clears the Blueprint override."
    },
    "speculativeMode": {
      "anyOf": [
        {
          "type": "string",
          "enum": [
            "off",
            "mtp"
          ]
        },
        {
          "type": "null"
        }
      ],
      "description": "Speculative decoding mode, limited to off or MTP; null clears the Blueprint override."
    },
    "mtpAutoTune": {
      "anyOf": [
        {
          "type": "boolean"
        },
        {
          "type": "null"
        }
      ],
      "description": "Automatically tune MTP drafting; relevant when speculativeMode is mtp. Null clears the Blueprint override."
    },
    "mtpDraftNMax": {
      "anyOf": [
        {
          "type": "integer",
          "minimum": 1,
          "maximum": 6
        },
        {
          "type": "null"
        }
      ],
      "description": "Maximum MTP draft tokens, relevant for MTP; null clears the Blueprint override."
    },
    "mtpDraftNMin": {
      "anyOf": [
        {
          "type": "integer",
          "minimum": 0,
          "maximum": 6
        },
        {
          "type": "null"
        }
      ],
      "description": "Minimum MTP draft tokens, no larger than the resolved maximum (2 when omitted); null clears the Blueprint override."
    },
    "mtpDraftPMin": {
      "anyOf": [
        {
          "type": "number",
          "minimum": 0,
          "maximum": 1
        },
        {
          "type": "null"
        }
      ],
      "description": "Minimum MTP draft probability from 0 through 1; null clears the Blueprint override."
    },
    "resetToDefaults": {
      "type": "boolean",
      "description": "Reset the webpage draft's required sampling settings and clear seed plus all optional advanced Blueprint overrides before applying supplied fields."
    }
  },
  "minProperties": 1,
  "additionalProperties": false
}