lytics-agent

name	lytics-agent
description	Top-level Lytics CDP agent - routes user intent to the appropriate skill for audience management, profile exploration, data integration, and schema browsing. Use when the user has a general Lytics question or request that doesn't clearly map to a specific skill.
metadata	{"arguments":"natural language request about Lytics CDP operations"}

Lytics Agent Router

Purpose

Top-level intent classifier and dispatcher for Lytics CDP operations. Analyzes the user's request and routes to the appropriate specialized skill.

Environment

Requires authenticated API access. See ../references/auth.md for credential resolution and pre-flight checks.

Intent Classification

Route based on keywords and intent:

Audience / Segment Operations

Triggers: "create audience", "build segment", "find users who", "people that", "customers who", "target audience", "audience of" Route to: audience-builder skill

Triggers: "list segments", "show audiences", "get segment", "segment details", "delete segment", "update segment" Route to: segment-manager skill

Audience Analytics / Snapshot

Triggers: "describe audience", "audience snapshot", "what does segment look like", "audience breakdown", "analyze segment", "audience profile", "audience analytics", "audience demographics", "who is in this segment" Route to: audience-snapshot skill

Audience Strategy / Advisor

Triggers: "best audience for", "optimize audience", "help me target", "who should I target", "improve my segment", "audience strategy", "what audience should", "right audience for", "how to reach", "better audience", "refine segment" Route to: audience-advisor skill

Profile Diagnostics

Triggers: "why is user in", "why isn't user in", "why not in segment", "diagnose profile", "what happened to user", "debug segment membership", "explain segment", "trace user" Route to: profile-investigator skill

Profile / Entity Exploration

Triggers: "look up user", "find profile", "who is", "show me user", "profile for", "entity lookup" Route to: profile-explorer skill

Schema / Data Model

Triggers: "what fields", "schema", "data model", "field types", "available data", "what data do we have" Route to: schema-manager skill

Schema Optimization

Triggers: "optimize schema", "unused fields", "schema cleanup", "field usage", "schema health", "improve schema", "inert fields", "missing mappings" Route to: schema-optimizer skill

Schema Discovery (for field mapping)

Triggers: "find field for", "what field matches", "map to field" Route to: schema-discovery skill

Integration Strategy / Advisor

Triggers: "connect to", "set up integration", "export to", "import from", "sync with", "send data to", "pull data from", "best way to connect", "how do I integrate" Route to: integration-advisor skill

Data Integration / Jobs (existing)

Triggers: "list jobs", "job status", "pause job", "resume job", "bounce job", "kill job", "job logs" Route to: job-manager skill

Campaign / Journey Builder

Triggers: "create flow", "build campaign", "set up journey", "welcome series", "nurture campaign", "email sequence", "multi-step campaign", "A/B test flow" Route to: campaign-flow-builder skill

Flows CRUD

Triggers: "list flows", "get flow", "delete flow", "flow status", "update flow", "pause flow" Route to: flow-manager skill

Export Debugging

Triggers: "why wasn't user exported", "export debug", "not showing up in", "didn't sync to", "missing from export", "why no export", "trace export", "export failed" Route to: export-debugger skill

Data Health

Triggers: "data health", "is my data flowing", "health check", "any data issues", "pipeline status", "what's broken", "data problems", "check my data", "how's my data" Route to: data-health-monitor skill

Streams / Events

Triggers: "stream", "events", "data flowing", "incoming data", "event log" Route to: stream-inspector skill

Connections / Auth

Triggers: "connection", "auth", "credentials", "provider" Route to: connection-manager skill

Webhook Templates

Triggers: "webhook template", "send to a webhook", "custom webhook destination", "Qualtrics webhook", "Slack webhook", "transform profile to webhook payload", "build a webhook payload", "webhook integration for", "send profiles to a custom URL" Route to: webhook-template-builder skill

Cross-Account Sync / Promotion

Triggers: "copy from sandbox to prod", "promote to prod", "sync to prod", "copy segment to another account", "sandbox to production", "promote audience", "promote flow", "copy schema between accounts", "move segment from X to Y", "sync from to ", "copy account settings", "promote settings", "sync settings between accounts", "copy idconfig", "sync account config", "promote account settings", "copy field rankings between accounts" Route to: account-sync skill

Multi-Intent Handling

Some requests combine multiple intents. Sequence skills in order:

• "Create an audience and export to Facebook"

  1. audience-builder skill -> create segment
  2. integration-advisor skill -> set up export with new segment

• "Find users in the high-value segment and show me one"

  1. segment-manager skill -> get segment details
  2. profile-explorer skill -> look up a member

• "What fields do we have for purchase data and build a segment from it"

  1. schema-discovery skill -> find purchase fields
  2. audience-builder skill -> build segment using discovered fields

• "Build a segment in sandbox and promote it to prod"

  1. audience-builder skill -> create segment in sandbox
  2. account-sync skill -> copy the new segment from sandbox to prod

Disambiguation

If the intent is unclear:

1. Ask a clarifying question
2. Suggest the most likely interpretation
3. Offer alternatives

Example: "Show me data" could mean:

• Schema/fields -> schema-manager
• Stream events -> stream-inspector
• Profile data -> profile-explorer

Ask: "Would you like to see your schema fields, recent streaming events, or look up a specific profile?"

Shared Conventions

All skills follow these conventions:

• Auth: See ../references/auth.md for credential resolution
• Base URL: See ../references/api-client.md for URL construction
• Reads: Execute immediately, display results
• Writes: Always use confirmation-gate pattern (summary + payload + approval)
• Errors: Parse API errors, suggest fixes, never silently fail