# 设置 API Key
export OPENAI_API_KEY=sk-xxx

# 启动 Hindsight
docker run --rm -it --pull always -p 8888:8888 -p 9999:9999 \
  -e HINDSIGHT_API_LLM_API_KEY=$OPENAI_API_KEY \
  -v $HOME/.hindsight-docker:/home/hindsight/.pg0 \
  ghcr.io/vectorize-io/hindsight:latest

export OPENAI_API_KEY=sk-xxx
export HINDSIGHT_DB_PASSWORD=choose-a-password
cd docker/docker-compose
docker compose up

docker run ... -e HINDSIGHT_API_LLM_PROVIDER=anthropic ...

# Python
pip install hindsight-client -U

# Node.js / TypeScript
npm install @vectorize-io/hindsight-client

from hindsight_client import Hindsight

client = Hindsight(base_url="http://localhost:8888")

# Retain: 存储记忆
client.retain(
    bank_id="my-bank",
    content="Alice works at Google as a software engineer"
)

# Recall: 检索记忆
results = client.recall(
    bank_id="my-bank",
    query="What does Alice do?"
)

# Reflect: 深度反思
insights = client.reflect(
    bank_id="my-bank",
    query="Tell me about Alice"
)

# 带上下文
client.retain(
    bank_id="my-bank",
    content="Alice got promoted to senior engineer",
    context="career update",
    timestamp="2025-06-15T10:00:00Z"
)

# 时间范围检索
results = client.recall(
    bank_id="my-bank",
    query="What happened in June?"
)

# 最简单的集成方式 - 替换现有 LLM Client
from hindsight import HindsightWrapper

# 替换你的 LLM Client
wrapper = HindsightWrapper(
    llm_provider="openai",
    llm_model="gpt-4o-mini"
)
# 之后所有的 LLM 调用都会自动存储和检索记忆

const { HindsightClient } = require('@vectorize-io/hindsight-client');

const client = new HindsightClient({
    baseUrl: 'http://localhost:8888'
});

// 存储记忆
await client.retain('my-bank', 'Alice loves hiking in Yosemite');

// 检索记忆
const results = await client.recall('my-bank', 'What does Alice like?');
console.log(results);

// 深度反思
const insights = await client.reflect('my-bank', 'Tell me about Alice');

client.retain(
    bank_id="my-bank",
    content="Alice works at Google"
)

results = client.recall(
    bank_id="my-bank",
    query="What does Alice do?"
)

insights = client.reflect(
    bank_id="my-bank",
    query="What should I know about Alice?"
)

# 存储带元数据的记忆
client.retain(
    bank_id="user-123",
    content="User prefers email over phone calls",
    metadata={"channel": "email", "user_id": "user-123"}
)

# 检索时过滤特定用户
results = client.recall(
    bank_id="user-123",
    query="User communication preferences?"
)

# 与 n8n、LangChain 等工作流集成
from hindsight import HindsightWrapper
from langchain.agents import initialize_agent

# 创建带记忆的 Agent
wrapper = HindsightWrapper(llm_provider="openai")
agent = initialize_agent(
    tools=[],
    llm=wrapper,
    memory=wrapper.get_memory()
)

# 单行启动
docker run --rm -it --pull always -p 8888:8888 -p 9999:9999 \
  -e HINDSIGHT_API_LLM_API_KEY=$OPENAI_API_KEY \
  ghcr.io/vectorize-io/hindsight:latest

version: '3.8'
services:
  postgres:
    image: pgvector/pgvector:pg16
    environment:
      POSTGRES_PASSWORD: ${HINDSIGHT_DB_PASSWORD}
    volumes:
      - postgres_data:/var/lib/postgresql/data

  hindsight:
    image: ghcr.io/vectorize-io/hindsight:latest
    ports:
      - "8888:8888"
      - "9999:9999"
    environment:
      HINDSIGHT_DB_PASSWORD: ${HINDSIGHT_DB_PASSWORD}
      HINDSIGHT_API_LLM_API_KEY: ${OPENAI_API_KEY}
    depends_on:
      - postgres

helm install hindsight oci://ghcr.io/vectorize-io/hindsight-chart \
  --set api.env.OPENAI_API_KEY=$OPENAI_API_KEY

pip install hindsight-all -U

import os
from hindsight import HindsightServer, HindsightClient

with HindsightServer(
    llm_provider="openai",
    llm_model="gpt-4o-mini",
    llm_api_key=os.environ["OPENAI_API_KEY"]
) as server:
    client = HindsightClient(base_url=server.url)
    client.retain(bank_id="my-bank", content="Alice works at Google")
    results = client.recall(bank_id="my-bank", query="Where does Alice work?")

# 在 Docker 中指定 Provider
docker run ... -e HINDSIGHT_API_LLM_PROVIDER=anthropic

# 或使用 Python SDK
wrapper = HindsightWrapper(
    llm_provider="anthropic",
    llm_model="claude-3-5-sonnet"
)

# 使用元数据隔离记忆
client.retain(
    bank_id="support-agent",
    content="Customer complained about billing",
    metadata={
        "user_id": "user-123",
        "channel": "email",
        "priority": "high"
    }
)

# 检索时过滤
results = client.recall(
    bank_id="support-agent",
    query="billing issues",
    metadata_filter={"channel": "email"}
)