# Steel Claude Code API Evaluation

**Category:** Browser
**Score:** 80 (B)

This is the agent-readable Devtool Arena evaluation page for Steel on Claude Code API.

## Summary

| Metric | Value |
|--------|-------|
| Status | completed |
| Overall score | 80 |
| Grade | B |
| Eval score | 72 |
| Discovery score | 100 |
| Cost | $0.25 |
| Runtime | 2m 30s |
| Tool calls | 13 |
| Errors | 2 |
| Tokens used | 263593 |
| Success | Yes |

## Agent-Readiness Checklist

| Signal | Evidence |
|--------|----------|
| Context7 | https://context7.com/steel-dev/steel-node |
| llms.txt | https://steel.dev/llms.txt |
| MCP server | https://github.com/steel-dev/steel-mcp-server |
| Typed SDK | https://github.com/steel-dev/steel-node |
| OpenAPI | https://api.steel.dev/sdk-openapi.json |
| Agent skills | https://github.com/steel-dev/skills |
| CLI | https://github.com/steel-dev/cli |

## Evaluation Prompt

````text

# Task
Using https://docs.steel.dev, complete the following task:

Using Steel, build me a Python script that:

1. Install the Steel Python package
2. Open a browser session
3. Navigate to "https://example.com"
4. Extract the page title and the main heading (h1) text
5. Take a screenshot and save as "screenshot.png" (if supported)
6. Close the session
7. Print results as JSON: {"url": "...", "title": "...", "heading_text": "...", "screenshot_saved": true/false}

Use the official Steel Python library — not raw HTTP calls. 

## Execution
After creating the script, run it to verify it works:
```bash
cd /home/daytona/app && python <your_script>.py
```

The script should print output to stdout. If there are errors, debug and fix them until it runs successfully.

````

## Grader Results

| Check | Passed | Weight | Score | Details |
|-------|--------|--------|-------|---------|
| Cost | Yes | 0 | 0 | $0.25 ($0.20–0.40, good) |
| Time | Yes | 0 | 0 | 150s (2–3min, good) |
| Efficiency | Yes | 0 | 0 | 13 tool calls (≤15, good) |
| Found Docs | Yes | 0 | — | Touched docs at docs.steel.dev: yes (7 matching tool calls) |
| Zero Errors | No | 0 | — | Tool outputs with errors/tracebacks: 2 |
| Syntax Valid | Yes | 0 | — | Code compiles/parses correctly |
| Created Files | Yes | 0 | — | Generated files: 1 (need ≥1) |
| Full Execution | Yes | 0 | — | Method: full_execution, Exit: 0 |

## Run Artifacts

| Artifact | Value |
|----------|-------|
| Conversation turns | 17 |
| Tool call traces | 13 |
| Generated files | 1 |
| Exit code | 0 |
| Completed at | 2026-09-23T05:32:49.641721+00:00 |

### Generated Files

- /home/daytona/app/steel_script.py

## Related Pages

- [Claude Code API leaderboard](/leaderboard/claudecode/api)
- [Compare Claude Code API companies](/leaderboard/claudecode/api/compare)
- [Agent Landscape](/leaderboard/discoverability)

Canonical URL: https://devtoolarena.com/claudecode/api/steel