update README.md

This commit is contained in:
2025-11-09 04:54:51 -05:00
parent 84564237d1
commit f869740f55
2 changed files with 67 additions and 2 deletions

View File

@@ -10,8 +10,73 @@ A DSPy module that wraps the OpenAI Codex SDK with a signature-driven interface.
- **Rich outputs** - Get typed results + execution trace + token usage - **Rich outputs** - Get typed results + execution trace + token usage
- **Multi-turn conversations** - Context preserved across calls - **Multi-turn conversations** - Context preserved across calls
- **Output field descriptions** - Automatically enhance prompts - **Output field descriptions** - Automatically enhance prompts
- **Modaic integration** - Deploy and share agents on the Modaic hub
## Installation ## Quick Start with Modaic
The fastest way to use CodexAgent is through [Modaic](https://docs.modaic.dev/), which allows you to pull pre-packaged agents from the hub.
### 1. Set up environment variables
Create a `.env` file with your API keys:
```bash
# Copy the example file
cp .env.example .env
# Edit .env with your keys
MODAIC_TOKEN="<YOUR_MODAIC_TOKEN>"
CODEX_API_KEY="<YOUR_CODEX_API_KEY>"
```
- Get your **MODAIC_TOKEN** from [Modaic dashboard](https://modaic.dev/)
- Get your **CODEX_API_KEY** (OpenAI API key) from [OpenAI platform](https://platform.openai.com/api-keys)
### 2. Install dependencies
```bash
pip install modaic dspy python-dotenv
```
### 3. Use the agent
```python
from modaic import AutoAgent
import dspy
from pydantic import BaseModel
from dotenv import load_dotenv
load_dotenv()
class Dependencies(BaseModel):
dependencies: list[str]
def main():
sig = dspy.Signature("message:str -> answer:Dependencies")
codex_agent = AutoAgent.from_precompiled(
"darinkishore/codex-agent",
config_options={
"signature": sig,
"working_directory": ".",
"model": "gpt-4o",
"codex_path_override": "/opt/homebrew/bin/codex" # Optional: customize codex binary path
},
)
result = codex_agent(message="check the pyproject.toml for this projects dependencies")
print(result.answer.dependencies) # Typed Pydantic output
print(result.trace) # Execution items (commands, files, etc.)
print(result.usage) # Token counts
if __name__ == "__main__":
main()
```
**Note**: Make sure the Codex CLI is installed and accessible. On macOS with Homebrew, it's typically at `/opt/homebrew/bin/codex`. Use `which codex` to find your path.
## Local Installation & Development
If you want to develop or modify the agent locally:
```bash ```bash
# Install dependencies # Install dependencies

View File

@@ -7,7 +7,7 @@ codex_agent = CodexAgent(CodexAgentConfig())
def main(): def main():
codex_agent.push_to_hub("darinkishore/codex-agent", with_code=True, commit_message="fix thread error") codex_agent.push_to_hub("darinkishore/codex-agent", with_code=True, commit_message="update README.md")
#result = codex_agent(message="What files are in this directory?") #result = codex_agent(message="What files are in this directory?")
#print(result.answer) # String response #print(result.answer) # String response
#print(result.trace) # Execution items (commands, files, etc.) #print(result.trace) # Execution items (commands, files, etc.)