Add example running a Claude Code coding agent in a Sandbox

13_sandboxes/sandbox_agent.py

@@ -0,0 +1,66 @@
+# ---
+# cmd: ["python", "13_sandboxes/sandbox_agent.py"]
+# pytest: false
+# ---
+
+# # Run Claude Code in a Modal Sandbox
+
+# This example demonstrates how to run Claude Code in a Modal
+# [Sandbox](https://modal.com/docs/guide/sandbox) to analyze a GitHub repository.
+# The Sandbox provides an isolated environment where the agent can safely execute code
+# and examine files.
+
+import modal
+
+app = modal.App.lookup("example-sandbox-agent", create_if_missing=True)
+
+# First we create a custom Image that has Claude Code installed.
+image = (
+    modal.Image.debian_slim(python_version="3.12")
+    .apt_install("curl", "git")
+    .run_commands(
+        "curl -fsSL https://claude.ai/install.sh | bash",
+    )
+    .env(
+        {
+            "PATH": "/root/.local/bin:$PATH",
+            "USE_BUILTIN_RIPGREP": "0",
+        }
+    )
+)
+
+with modal.enable_output():
+    sandbox = modal.Sandbox.create(app=app, image=image)
+print(f"Sandbox ID: {sandbox.object_id}")
+
+# Next we'll clone a repository that Claude Code will work on.
+repo_url = "https://github.com/modal-labs/modal-examples"
+git_ps = sandbox.exec("git", "clone", repo_url, "/repo")
+git_ps.wait()
+print(f"Cloned '{repo_url}' into /repo.")
+
+# Finally we'll run Claude Code to analyze the repository.
+claude_cmd = [
+    "claude",
+    "-p",
+    "Summarize what this repository is about. Don't modify any files.",
+]
+
+print("\nRunning command:", claude_cmd)
+
+claude_ps = sandbox.exec(
+    *claude_cmd,
+    pty=True,  # Adding a PTY is important, since Claude requires it!
+    secrets=[
+        modal.Secret.from_name("anthropic-secret", required_keys=["ANTHROPIC_API_KEY"])
+    ],
+    workdir="/repo",
+)
+claude_ps.wait()
+
+print("\nAgent stdout:\n")
+print(claude_ps.stdout.read())
+
+stderr = claude_ps.stderr.read()
+if stderr != "":
+    print("Agent stderr:", stderr)