dogebi.github.io WebMCP Tools, Schemas & Agent Capabilities

dogebi.github.io

live Unverified (seed import) AI & Agents 8 tools imperative implementation

Offline Preact and WebGL2 architecture atlas for DeepSeek V4.1 Flash: causal encoder-decoder, CSA2, Engram, DSpark, native FP4 and supplied NVFP4 recipe, byte-proportional solids and local…

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

Answer 6 · Act 2 · Transact 0 · Score: Not yet scored

Tool inspector

search_tensors

answer imperative

Search weight tensors by name substring, category, or storage format (e.g. "experts.w1", cat "attention", format "fp4"). Returns matching tensor names with shapes, formats, parameter counts and the module/layer they belong to.

Page: /dsv41-atlas/

Input schema

Property	Type	Required	Description
`query`	string	no	Substring to match against the tensor name (case-insensitive). Leave empty to match all.
`category`	string	no	Exact category key, e.g. routed, attention, vision, engram, shared, vocab, dspark, other.
`format`	string	no	Storage format, e.g. fp4, fp8, bf16.
`module`	string	no	Restrict to one module id, e.g. embed or L0.
`limit`	integer	no	Max rows to return (default 20, max 200).

Raw JSON schema

{
  "type": "object",
  "properties": {
    "query": {
      "type": "string",
      "description": "Substring to match against the tensor name (case-insensitive). Leave empty to match all."
    },
    "category": {
      "type": "string",
      "description": "Exact category key, e.g. routed, attention, vision, engram, shared, vocab, dspark, other."
    },
    "format": {
      "type": "string",
      "description": "Storage format, e.g. fp4, fp8, bf16."
    },
    "module": {
      "type": "string",
      "description": "Restrict to one module id, e.g. embed or L0."
    },
    "limit": {
      "type": "integer",
      "description": "Max rows to return (default 20, max 200)."
    }
  },
  "additionalProperties": false
}