twilio-customer-memory

Store and retrieve customer context using Twilio Conversation Memory APIs.

5.3k|765|Updated Mar 4, 2026
One-click install
npx skills add https://github.com/openai/plugins --skill twilio-customer-memory
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: twilio-customer-memory
Source: https://github.com/openai/plugins/tree/main/plugins/twilio-developer-kit/skills/twilio-customer-memory
Command: npx skills add https://github.com/openai/plugins --skill twilio-customer-memory

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires requests.

What problem does it solve?

AI agents and human agents lose customer context between sessions and channels, forcing customers to repeat themselves. This Skill shows how to persist observations, traits, and conversation summaries in Twilio Conversation Memory and recall them before responding.

Core Features & Use Cases

  • Memory Store Provisioning: Create and manage Memory Stores on memory.twilio.com, including multi-tenant (ISV) patterns with one store per client.
  • Profiles, Traits, and Observations: Create customer profiles with trait groups, look up profiles by phone number, and post batched observations from custom extraction pipelines.
  • Semantic Recall: Retrieve relevant observations and summaries via hybrid lexical and semantic search, using a conversation ID, custom query, or chronological mode.
  • Use Case: A voice AI agent on ConversationRelay looks up the caller's profile at call start, injects recalled history into the system prompt, and writes a call summary observation when the call ends.

Quick Start

Ask the agent to create a Twilio Memory Store and show how to recall customer context for a profile before responding.

Frequently Asked Questions about twilio-customer-memory

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I give an AI agent persistent memory of customer conversations?

Use Twilio Conversation Memory to store observations and traits per customer profile, then call the Recall endpoint before responding. Recall runs hybrid lexical and semantic search and returns relevant context to inject into the LLM prompt.

How do I look up a Twilio customer profile by phone number?

Post to the Profiles/Lookup endpoint with idType set to phone and the caller's number. The response returns matching profiles, which is useful for inbound voice calls where only the caller number is known.

Does Twilio Conversation Memory extract observations automatically?

Yes, when conversations flow through Conversation Orchestrator, observations and summaries are extracted automatically at conversation close or inactivity. Custom pipelines outside Orchestrator must post observations directly to the Observations endpoint.

Why are my Twilio memory observations not appearing in Recall results?

Recall has eventual consistency, so allow about two seconds after writing before querying. Also verify the Memory Store is correctly linked in the Conversation Orchestrator config, since misconfigured linkages silently skip extraction.

What are the limits of Twilio Conversation Memory?

Accounts are limited to 15 Memory Stores, Recall returns at most 20 observations per query, and batch writes accept up to 10 observations per request. Profile deletion is permanent and cannot be recovered.