32 lines
934 B
Markdown
32 lines
934 B
Markdown
# LangGraph Agent with Conversation Memory and User Confirmation
|
||
|
||
## Installation
|
||
|
||
```bash
|
||
python -m venv .venv
|
||
source .venv/bin/activate # on Windows use `.venv\Scripts\activate`
|
||
pip install -r requirements.txt
|
||
```
|
||
|
||
## Environment Variables
|
||
|
||
The agent uses the default OpenAI endpoint via `langchain`. If you want to override, set:
|
||
|
||
- `OPENAI_API_KEY` – your OpenAI API key.
|
||
- `OPENAI_BASE_URL` – custom base URL (e.g., for local Ollama).
|
||
|
||
## Running the Agent
|
||
|
||
```bash
|
||
python agent.py
|
||
```
|
||
|
||
The script starts with an empty conversation history. The graph will:
|
||
1. Retrieve memory (currently just echoes the existing history).
|
||
2. Invoke a placeholder agent that would normally process the conversation.
|
||
3. Prompt you to confirm each tool call via Rich console output.
|
||
4. Continue execution after confirmation.
|
||
|
||
## Extending
|
||
|
||
Replace `agent_node` with real LangChain logic, add tools, and integrate an actual LLM model as needed. |