Back to Prompt Library
implementation

Implement Tactical Commander with GPT-5 Pro and Tool Use

Inspect the original prompt language first, then copy or adapt it once you know how it fits your workflow.

Linked challenge: Battlefield AI Decision Support Agent

Format
Code-aware
Lines
5
Sections
1
Linked challenge
Battlefield AI Decision Support Agent

Prompt source

Original prompt text with formatting preserved for inspection.

5 lines
1 sections
No variables
1 code block
Implement the 'Tactical Commander' agent using the Claude Agents SDK. This agent should primarily use GPT-5 Pro for high-level strategic reasoning and planning. Define custom tools for this agent to interact with 'Featherless AI' (e.g., to request fast object detection from a drone feed) and 'Replicate' (e.g., to invoke a specialized weather prediction model). Show the Claude Agents SDK agent definition including tool invocation. ```python
import anthropic # Assuming an Anthropic client setup and GPT-5 Pro tool definition
# For demonstration, a simplified 'tool' interface def call_gpt5_pro(prompt: str) -> str: # Placeholder for actual GPT-5 Pro API call return f"GPT-5 Pro response to: {prompt[:50]}..." def get_object_detection_from_featherless(image_url: str) -> str: # Placeholder for Featherless AI integration return f"Featherless AI detected objects in {image_url}." def get_weather_prediction_from_replicate(coords: str) -> str: # Placeholder for Replicate integration return f"Replicate weather prediction for {coords}." # In Claude Agents SDK, tool definitions would be more formal
# This is a conceptual snippet for prompt guidance class TacticalCommanderAgent: def __init__(self, client: anthropic.Anthropic): self.client = client self.tools = { "plan_strategy": call_gpt5_pro, "detect_objects": get_object_detection_from_featherless, "predict_weather": get_weather_prediction_from_replicate, } async def process_threat(self, threat_data: str): # Logic for the agent to call tools and use Claude models pass # ... (actual Claude Agents SDK implementation would follow)
```

Adaptation plan

Keep the source stable, then change the prompt in a predictable order so the next run is easier to evaluate.

Keep stable

Hold the task contract and output shape stable so generated implementations remain comparable.

Tune next

Update libraries, interfaces, and environment assumptions to match the stack you actually run.

Verify after

Test failure handling, edge cases, and any code paths that depend on hidden context or secrets.