template.md

---
# Agent Card
agent_name: ""
version: ""
developers: ""
license: ""
framework: ""
repository: ""
models:
  - ""
tags:
  - agent
---

# {{ agent_name }}

> One-line description of what this agent does.

## Agent Details

- **Name**: {{ agent_name }}
- **Version**: {{ version }}
- **Developers**: {{ developers }}
- **Framework**: [{{ framework }}]({{ repository }})
- **License**: {{ license }}

## Architecture

<!-- Describe how the agent works. What makes it different from other agents? -->

### Planning
<!-- How does the agent decide what to do next? (e.g., ReAct, chain-of-thought, tree search) -->

### Memory
<!-- What context does the agent maintain across steps? (e.g., conversation history, scratchpad, retrieved docs) -->

### Tool Use
<!-- How does the agent interact with tools/actions? (e.g., function calling, code generation, CLI) -->

### Error Recovery
<!-- How does the agent handle failures? (e.g., retry, replan, fallback) -->

### Subagents
<!-- Does the agent delegate to other agents? If so, which and when? -->

## Models

<!-- Which LLMs does the agent use, and for what purpose? -->

| Role | Model | Purpose |
|------|-------|---------|
| Main | | Primary reasoning and action selection |

## Supported Environments

<!-- What domains, benchmarks, or task types does this agent handle? -->

-

## Evaluation Results

<!-- Summary of benchmark performance. Link to full results on the leaderboard. -->

| Benchmark | Score | Avg Cost | Avg Steps |
|-----------|-------|----------|-----------|
| | | | |

**Full results**: [Open Agent Leaderboard](https://huggingface.co/spaces/open-agent-leaderboard/leaderboard)

## Cost Profile

<!-- Typical cost per task, and what drives cost (model calls, retries, long sessions). -->

## Limitations

<!-- Known failure modes, unsupported scenarios, edge cases. Be honest -- this helps users make informed decisions. -->

## How to Run

<!-- Installation and basic usage instructions. -->

```bash
# Installation

# Usage
```