dspy-skills

🥈Silver

DSPy Skills Collection automates LLM programming and prompt optimization for operations teams. It enables systematic RAG pipeline development, connecting to Python-based workflows and Claude agents.

6340Updated 1mo ago

Intermediate30min to implementautomation

Saves ~100 min per use

Quick InstallView Source

git clone https://github.com/OmidZamani/dspy-skills.git

Works with:

Claude

Overview

About This Skill

DSPy Skills Collection automates LLM programming and prompt optimization for operations teams. It enables systematic RAG pipeline development, connecting to Python-based workflows and Claude agents.

How to Use

["Install DSPy and dependencies: `pip install dspy-ai pydantic`","Define your task requirements in the prompt template, including [TASK_DESCRIPTION] and [OPERATIONS_TEAM]","Connect to your Python workflow by replacing [PYTHON_WORKFLOW] with your module imports or API calls","Specify output format in [FORMAT] (e.g., JSON, CSV, or structured Python objects)","Test edge cases in [EDGE_CASES] (e.g., invalid queries, missing data) and verify logging in [LOGGING_SYSTEM]","Tip: Use DSPy's `teleprompt` module to fine-tune the pipeline with your organization's data patterns"]

Setup & Installation

Quick Install

No install command available. Check the GitHub repository for manual installation instructions.

Alternative Install (Git Clone)

git clone https://github.com/OmidZamani/dspy-skills

Requirements

Claude Code or compatible AI agent
Works with: Claude

Quick Start Guide

Install the Skill

Copy the install command above and run it in your terminal.

Open Your AI Agent

Launch Claude Code, Cursor, or your preferred AI coding agent.

Try It Out

Use the prompt template or examples below to test the skill.

Customize

Adapt the skill to your specific use case and workflow.

Usage Examples

Prompt Template

Use DSPy to optimize the [TASK_DESCRIPTION] for [OPERATIONS_TEAM]. Implement a RAG pipeline that connects to [PYTHON_WORKFLOW] and outputs structured data in [FORMAT]. Include error handling for [EDGE_CASES] and log results to [LOGGING_SYSTEM].

Example Output

```python
import dspy
from dspy.teleprompt import BootstrapFewShot

# Define the task
class OptimizedRAG(dspy.Module):
    def __init__(self):
        super().__init__()
        self.retrieve = dspy.Retrieve(k=3)
        self.generate = dspy.ChainOfThought("question -> answer")

    def forward(self, question):
        context = self.retrieve(question)
        return self.generate(context=context, question=question)

# Compile with DSPy
train = [dspy.Example(question="What are the Q3 sales targets for the EMEA region?", answer="€12.5M").with_inputs("question")]
optimizer = BootstrapFewShot(
    metric=lambda example, pred: dspy.evaluate.answer_exact_match(example, pred),
    max_bootstrapped_demos=4,
    max_labeled_demos=16
)
compiled_pipeline = optimizer.compile(OptimizedRAG(), trainset=train)

# Execute in Python workflow
result = compiled_pipeline(question="What are the Q3 sales targets for the EMEA region?")
print(f"Optimized Answer: {result.answer}")
```

**Execution Log:**
- Retrieved 3 context chunks from internal sales database
- Applied Chain of Thought reasoning to synthesize answer
- Validated against Q3 target dataset (accuracy: 98.7%)
- Output: `Optimized Answer: €12.5M`
- Logging: Stored in `operations_logs/sales_targets.log` with timestamp 2024-05-15T14:32:11Z