threadline-sdk

The persistent context layer for AI agents.

Every agent you build starts from zero. Threadline changes that.

Install

npm install threadline-sdk

Usage

import { Threadline } from "threadline-sdk"

const tl = new Threadline({ apiKey: process.env.THREADLINE_KEY })

// Before your AI call — inject user context
const { injectedPrompt, cacheHint } = await tl.inject(userId, "You are a helpful assistant.")

// After your AI call — update user context
await tl.update({ userId, userMessage, agentResponse })

That's it. Your agent now remembers every user, across every session.

Why Threadline?

Think of it like OAuth — but for context. You don't build auth from scratch. You shouldn't build memory from scratch either.

• One SDK, any LLM (OpenAI, Anthropic, Gemini, Mistral)
• User-owned context — users can view and delete their data
• < 50ms context retrieval via Redis
• Privacy-first by design

API

tl.inject(userId, basePrompt)

Returns { injectedPrompt, cacheHint? }. Use injectedPrompt as your system message. When cacheHint.recommended is true, pass cacheHint.openaiParam as extra_body on supported OpenAI models for 24h prompt cache retention (~50% lower cached input token cost).

tl.update({ userId, userMessage, agentResponse })

Extracts and stores context updates from a conversation turn.

Get started

1. Sign up at threadline.to
2. Create an agent and get your API key
3. Drop in the two lines above

Links

• Documentation
• Quickstart
• Examples