Bring production logs and AI agent traces straight into your DevRev Agents & Workflows.

Operations for querying MELTS service logs with attribute filters and retrieving AI agent conversation traces with structured parsing. Designed for workflow automation and agent skills.

Overview

MELTS Observability Operations gives DevRev workflows and agents direct access to production observability data — service logs and AI agent execution traces — without manual API calls or context switching. Built for incident triage, agent debugging, and automated investigation flows.

Features

  • Get Service Logs — Query MELTS log datasets by service name, time range, attribute filters (@key:value syntax), and body text search. Returns structured log entries with timestamps, severity, and attributes.
  • Get Agent Traces — Extract AI agent execution traces from conversation, ticket, or DM timeline comments. Parses raw spans into structured sessions showing system context, available tools, queries, steps, and token usage.
  • Workflow-ready — Both operations work as workflow steps and as agent skills, enabling automated debugging and self-inspection patterns.
  • Flexible filtering — Partial or exact match modes, multi-service queries, comma-separated service lists, configurable limits up to 2000 logs.

Use cases

  • Investigate service errors within a workflow triggered by an alert
  • Let agents self-inspect their own execution traces for debugging
  • Pull logs for a specific time window during incident triage
  • Analyze agent tool usage and decision patterns across conversations
  • Build automated postmortem workflows that gather logs and traces by incident time window