Personal Assistant

Build a personal AI assistant that helps with daily tasks and remembers context across conversations.

System prompt

You are my personal assistant. You help with daily work tasks, research,
and decision-making.

When I tell you a preference or rule, save it using your memory tool so
you can recall it in future conversations. When starting a new conversation,
check your memory for relevant context about me and my work.

MCP server: Persistent Memory

The assistant needs an MCP server that provides memory tools — save and recall information across sessions. The MCP server would expose tools like:

• save_memory — stores a fact, preference, or note with a topic tag
• search_memory — retrieves relevant memories by semantic search

This could be backed by a vector database (pgvector, Pinecone) or a simple key-value store. The agent calls these tools based on the system prompt instructions.

API example

curl -s -X POST https://zhenfy.ai/api/agent/v1/agents \
  -H "Authorization: Bearer $TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "name": "My Assistant",
    "system_prompt": "You are my personal assistant. Save preferences using your memory tool. Check memory at the start of each conversation.",
    "tools": [
      {"url": "https://memory.example.com/mcp"}
    ]
  }'

Example conversation

You: I'm a DevOps engineer. I work mostly with Kubernetes and AWS.
Agent: Got it — I've saved that to memory.

You: Always respond in bullet points when giving me steps.
Agent: Noted — saved as a preference.

--- (new session) ---

You: What do you know about me?
Agent: From my memory:
  - You're a DevOps engineer
  - You work with Kubernetes and AWS
  - You prefer bullet points for step-by-step instructions