Skip to main content

​
Match SDK

Policy-based compute matching and route planning

​
Operations

MethodDescription
explainEvaluate policy against compute/model meta
computeMatch compute for model
planRoutePlan a route (match + resolve endpoint)

​
Generated Reference

​
Overview

Policy-based compute matching and route planning

​
Available Operations

  • explain - Evaluate policy against compute/model meta
  • compute - Match compute for model
  • planRoute - Plan a route (match + resolve endpoint)

​
explain

Evaluate a policy against compute and model metadata, returning a detailed explanation of whether the compute node is allowed and the reasons for the decision. Useful for debugging policy mismatches.

​
Example Usage

import { RaijinLabsLucidAi } from "raijin-labs-lucid-ai";

const raijinLabsLucidAi = new RaijinLabsLucidAi();

async function run() {
  const result = await raijinLabsLucidAi.match.explain({});

  console.log(result);
}

run();

​
Standalone function

The standalone function version of this method: 
import { RaijinLabsLucidAiCore } from "raijin-labs-lucid-ai/core.js";
import { matchExplain } from "raijin-labs-lucid-ai/funcs/matchExplain.js";

// Use `RaijinLabsLucidAiCore` for best tree-shaking performance.
// You can create one instance of it to use across an application.
const raijinLabsLucidAi = new RaijinLabsLucidAiCore();

async function run() {
  const res = await matchExplain(raijinLabsLucidAi, {});
  if (res.ok) {
    const { value: result } = res;
    console.log(result);
  } else {
    console.log("matchExplain failed:", res.error);
  }
}

run();

​
Parameters

ParameterTypeRequiredDescription
requestoperations.LucidMatchExplainRequest:heavy_check_mark:The request object to use for the request.
optionsRequestOptions:heavy_minus_sign:Used to set various options for making HTTP requests.
options.fetchOptionsRequestInit:heavy_minus_sign:Options that are passed to the underlying HTTP request. This can be used to inject extra headers for examples. All Request options, except method and body, are allowed.
options.retriesRetryConfig:heavy_minus_sign:Enables retrying HTTP requests under certain failure conditions.

​
Response

Promise<operations.LucidMatchExplainResponse>

​
Errors

Error TypeStatus CodeContent Type
errors.ErrorResponse400application/json
errors.ErrorResponse500application/json
errors.RaijinLabsLucidAiDefaultError4XX, 5XX*/*

​
compute

x402-gated with dynamic pricing when X402_ENABLED=true.

​
Example Usage

import { RaijinLabsLucidAi } from "raijin-labs-lucid-ai";

const raijinLabsLucidAi = new RaijinLabsLucidAi();

async function run() {
  const result = await raijinLabsLucidAi.match.compute({
    body: {
      modelMeta: {
        schemaVersion: "1.0",
        modelPassportId: "ppt_model_7xK9mQ2v",
        format: "safetensors",
        runtimeRecommended: "vllm",
        contextLength: 32768,
        requirements: {
          minVramGb: 16,
        },
      },
      policy: {
        policyVersion: "1.0",
        allowRegions: [
          "us-east-1",
        ],
        latency: {
          p95MsBudget: 500,
        },
      },
      computeCatalog: [
        {
          schemaVersion: "1.0",
          computePassportId: "ppt_compute_Xn5vR2kJ",
          providerType: "cloud",
          regions: [
            "us-east-1",
          ],
          hardware: {
            gpu: "NVIDIA-A100-40GB",
            vramGb: 40,
          },
          runtimes: [],
          endpoints: {
            inferenceUrl: "https://ironclad-puritan.biz/",
          },
        },
      ],
    },
  });

  console.log(result);
}

run();

​
Standalone function

The standalone function version of this method: 
import { RaijinLabsLucidAiCore } from "raijin-labs-lucid-ai/core.js";
import { matchCompute } from "raijin-labs-lucid-ai/funcs/matchCompute.js";

// Use `RaijinLabsLucidAiCore` for best tree-shaking performance.
// You can create one instance of it to use across an application.
const raijinLabsLucidAi = new RaijinLabsLucidAiCore();

async function run() {
  const res = await matchCompute(raijinLabsLucidAi, {
    body: {
      modelMeta: {
        schemaVersion: "1.0",
        modelPassportId: "ppt_model_7xK9mQ2v",
        format: "safetensors",
        runtimeRecommended: "vllm",
        contextLength: 32768,
        requirements: {
          minVramGb: 16,
        },
      },
      policy: {
        policyVersion: "1.0",
        allowRegions: [
          "us-east-1",
        ],
        latency: {
          p95MsBudget: 500,
        },
      },
      computeCatalog: [
        {
          schemaVersion: "1.0",
          computePassportId: "ppt_compute_Xn5vR2kJ",
          providerType: "cloud",
          regions: [
            "us-east-1",
          ],
          hardware: {
            gpu: "NVIDIA-A100-40GB",
            vramGb: 40,
          },
          runtimes: [
            {
              name: "tensorrt",
            },
          ],
          endpoints: {
            inferenceUrl: "https://ironclad-puritan.biz/",
          },
        },
      ],
    },
  });
  if (res.ok) {
    const { value: result } = res;
    console.log(result);
  } else {
    console.log("matchCompute failed:", res.error);
  }
}

run();

​
Parameters

ParameterTypeRequiredDescription
requestoperations.LucidMatchRequest:heavy_check_mark:The request object to use for the request.
optionsRequestOptions:heavy_minus_sign:Used to set various options for making HTTP requests.
options.fetchOptionsRequestInit:heavy_minus_sign:Options that are passed to the underlying HTTP request. This can be used to inject extra headers for examples. All Request options, except method and body, are allowed.
options.retriesRetryConfig:heavy_minus_sign:Enables retrying HTTP requests under certain failure conditions.

​
Response

Promise<operations.LucidMatchResponse>

​
Errors

Error TypeStatus CodeContent Type
errors.X402PaymentRequiredError402application/json
errors.ErrorResponse422application/json
errors.ErrorResponse500application/json
errors.RaijinLabsLucidAiDefaultError4XX, 5XX*/*

​
planRoute

Perform compute matching and resolve an executable inference endpoint in a single call. Returns the matched compute node, model, endpoint URL, runtime, policy hash, and fallback options.

​
Example Usage

import { RaijinLabsLucidAi } from "raijin-labs-lucid-ai";

const raijinLabsLucidAi = new RaijinLabsLucidAi();

async function run() {
  const result = await raijinLabsLucidAi.match.planRoute({});

  console.log(result);
}

run();

​
Standalone function

The standalone function version of this method: 
import { RaijinLabsLucidAiCore } from "raijin-labs-lucid-ai/core.js";
import { matchPlanRoute } from "raijin-labs-lucid-ai/funcs/matchPlanRoute.js";

// Use `RaijinLabsLucidAiCore` for best tree-shaking performance.
// You can create one instance of it to use across an application.
const raijinLabsLucidAi = new RaijinLabsLucidAiCore();

async function run() {
  const res = await matchPlanRoute(raijinLabsLucidAi, {});
  if (res.ok) {
    const { value: result } = res;
    console.log(result);
  } else {
    console.log("matchPlanRoute failed:", res.error);
  }
}

run();

​
Parameters

ParameterTypeRequiredDescription
requestoperations.LucidRouteRequest:heavy_check_mark:The request object to use for the request.
optionsRequestOptions:heavy_minus_sign:Used to set various options for making HTTP requests.
options.fetchOptionsRequestInit:heavy_minus_sign:Options that are passed to the underlying HTTP request. This can be used to inject extra headers for examples. All Request options, except method and body, are allowed.
options.retriesRetryConfig:heavy_minus_sign:Enables retrying HTTP requests under certain failure conditions.

​
Response

Promise<operations.LucidRouteResponse>

​
Errors

Error TypeStatus CodeContent Type
errors.ErrorResponse422application/json
errors.ErrorResponse500application/json
errors.RaijinLabsLucidAiDefaultError4XX, 5XX*/*