Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.trodo.ai/docs/llms.txt

Use this file to discover all available pages before exploring further.

One call to wrapAgent produces one run in your dashboard. Give it an agent name and an async function; the SDK generates a run_id, opens a tracing context, runs your function, captures every child span, and flushes the result.

​
Example

import trodo from 'trodo-node';

trodo.init({ siteId: process.env.TRODO_SITE_ID });

const { result, runId } = await trodo.wrapAgent(
  'support-agent',                         // agentName  (required)
  async (run) => {
    run.setInput({ question: 'refund policy?' });
    run.setMetadata({ env: 'prod', version: '1.2.0' });

    try {
      const answer = await askLlm(...);   // spans captured inside the callback
      run.setOutput({ answer });           // run.status becomes 'ok'
      return answer;
    } catch (err) {
      run.setErrorSummary(err.message);   // run.status becomes 'error'
      throw err;
    }
  },
  { distinctId: 'user-42', conversationId: 'thread-abc' },
);

​
What you set vs what’s rolled up

You set on the RunHandleAuto-rolled from spans
agentName, distinctId, conversationIdduration_ms (callback start → end)
setInput / setOutput / setMetadatatotal_tokens_in, total_tokens_out
setErrorSummarytotal_cost
—span_count, tool_count, error_count
—status (ok / error / running)
Duration, tokens, cost, and status are computed for you — you never write them. total_tokens_in / total_tokens_out / total_cost are summed across every kind='llm' span inside the run.

​
Options

Option (Node / Python)Purpose
agentNameRequired. Free-text identifier — "support-bot", "ticket-triage-v2".
distinctId / distinct_idYour user id. Lets you filter runs per user.
conversationId / conversation_idGroups runs from the same chat thread.
parentRunId / parent_run_idMarks this run as a sub-agent. See Recipes → Sub-agents.
metadataFree-form JSON tags. Same as calling setMetadata up-front.

​
Next