Built-in Code Execution & Web Fetch — Code Execution & Web Tools

[ ABORT TO HUD ]
SEQ. 1

Built-in Code Execution & Web Fetch

⚙️ Code Execution & Web Tools⏱ 20 min⭐ 700 BASE XP⌨ HANDS-ON LAB

GA Code Execution, Web Fetch & Agent Skills

Anthropic's GA Code Execution tool (code_execution_20250825 / code_execution_20260120, no beta header required) gives Claude a stateful, sandboxed Bash + Python container to execute calculations, analyze Files API uploads, and generate downloadable artifacts.

Combining Code Execution, Web Search & Web Fetch

You can combine code_execution_20250825, web_search_20250305, and Web Fetch (web_fetch_20250910 / web_fetch_20260209 with dynamic sandbox filtering) alongside container.skills in a single Messages API request—allowing Claude to retrieve full URLs, filter noisy HTML in Python before it enters context, and return verified metrics.

⌨ HANDS-ON LABRun Real Code in Claude's Sandbox
⭐ +200 XP

Stop asking Claude to 'imagine' computation results. Enable the built-in code execution tool so it actually runs Python in the sandbox - then combine it with web search in one request.

1Enable the GA code execution tool on claude-sonnet-5-5 and ask for a real computation.
2Combine built-in tools in one request: web_search to fetch live data and code_execution to process it.
lab-sandbox — simulated environment
INFINITY LAB SANDBOX v2.6 — simulated shell
Type the command for the current objective. Helpers: "hint", "solution", "clear".
$
OBJECTIVE 1 / 2 — type "hint" if stuck
SYNAPSE VERIFICATION
QUERY 1 // 1
How does built-in code execution improve mathematical and data analysis accuracy?
It estimates outputs faster
It executes real Python code in an isolated sandbox and reads stdout/stderr directly
It disables model thinking
It eliminates input tokens