Execution API

Run shell commands and scripts inside sandboxes via the Execution API. Execute any process, stream output in real time, and capture exit codes.

Endpoint

Method	Path	Description
POST	/v1/sandboxes/:id/execute	Execute a command inside a sandbox

---

POST /v1/sandboxes/:id/execute

Run a command or script inside a running sandbox. The sandbox must be in the running state.

Basic Request

curl -X POST https://api.box.runra.dev/v1/sandboxes/sb_x7k2m9p4q1w3v5b8n/execute \
  -H "Authorization: Bearer $RUNRA_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "command": "echo Hello from sandbox && node --version"
  }'

Response — 200 OK

{
  "stdout": "Hello from sandbox\nv22.12.0\n",
  "stderr": "",
  "exitCode": 0,
  "durationMs": 245,
  "signal": null
}

Request Body

Field	Type	Required	Default	Description
command	string	Yes	—	Shell command to execute (bash)
cwd	string	No	/workspace	Working directory for the command
env	object	No	{}	Additional environment variables (merged with sandbox env)
timeoutMs	integer	No	60000	Command timeout in ms (max 300000 = 5 min)
stdin	string	No	—	Text to pipe to the command's stdin

Response Fields

Field	Type	Description
stdout	string	Standard output from the command
stderr	string	Standard error from the command
exitCode	integer	Process exit code. 0 = success, non-zero = error
durationMs	integer	Wall-clock execution time in milliseconds
signal	string | null	Signal that killed the process, or null if exited normally

---

Advanced Usage

Environment Variables

Pass environment variables per execution. These are merged with the sandbox-level env:

curl -X POST https://api.box.runra.dev/v1/sandboxes/sb_x7k2m9p4q1w3v5b8n/execute \
  -H "Authorization: Bearer $RUNRA_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "command": "echo $MY_VAR && echo $CI",
    "env": {
      "MY_VAR": "custom value",
      "CI": "true"
    }
  }'

Execution-level env vars take precedence over sandbox-level env vars with the same name.

Working Directory

Specify a different working directory:

curl -X POST https://api.box.runra.dev/v1/sandboxes/sb_x7k2m9p4q1w3v5b8n/execute \
  -H "Authorization: Bearer $RUNRA_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "command": "cat package.json",
    "cwd": "/workspace/my-project"
  }'

Timeout

Set per-command timeouts. The default is 60 seconds, maximum is 5 minutes:

curl -X POST https://api.box.runra.dev/v1/sandboxes/sb_x7k2m9p4q1w3v5b8n/execute \
  -H "Authorization: Bearer $RUNRA_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "command": "npm run build",
    "timeoutMs": 120000
  }'

If the command exceeds the timeout, it receives SIGKILL and the response returns:

{
  "stdout": "Starting build...\n",
  "stderr": "",
  "exitCode": null,
  "durationMs": 120001,
  "signal": "SIGKILL"
}

stdin

Pipe input to a command:

curl -X POST https://api.box.runra.dev/v1/sandboxes/sb_x7k2m9p4q1w3v5b8n/execute \
  -H "Authorization: Bearer $RUNRA_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "command": "grep error",
    "stdin": "info: starting\ninfo: processing\nerror: something broke\ninfo: done\n"
  }'

Response:

{
  "stdout": "error: something broke\n",
  "stderr": "",
  "exitCode": 0,
  "durationMs": 12,
  "signal": null
}

---

Streaming Output

For long-running commands, stream stdout and stderr in real time using Server-Sent Events (SSE).

Request with Streaming

curl -X POST https://api.box.runra.dev/v1/sandboxes/sb_x7k2m9p4q1w3v5b8n/execute \
  -H "Authorization: Bearer $RUNRA_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Accept: text/event-stream" \
  -d '{
    "command": "for i in 1 2 3; do echo Line $i; sleep 1; done"
  }'

Streaming Response

event: stdout
data: Line 1

event: stdout
data: Line 2

event: stdout
data: Line 3

event: done
data: {"exitCode":0,"durationMs":3012,"signal":null}

Event Types

Event	Payload	Description
stdout	string	A line of stdout output
stderr	string	A line of stderr output
done	JSON object	Command finished (includes exitCode, durationMs, signal)
error	JSON object	Execution error

---

SDK Usage

TypeScript

import { Runra } from "@runra/sandbox";

const client = new Runra();

const sandbox = await client.sandboxes.create({ image: "node:22" });

// Simple execution
const result = await sandbox.exec("npm install && npm run build");
console.log(result.exitCode); // 0
console.log(result.stdout);

// With options
const result = await sandbox.exec("npm test", {
  cwd: "/workspace/packages/core",
  env: { CI: "true" },
  timeoutMs: 120_000,
});

Python

from runra import Runra

client = Runra()
sandbox = client.sandboxes.create(image="node:22")

result = sandbox.exec("ls -la /workspace")
print(result.stdout)
print(f"Exit code: {result.exit_code}")

---

Exit Codes

Exit Code	Meaning
0	Success
1–127	Application-specific error
137	Killed by SIGKILL (timeout or OOM)
143	Killed by SIGTERM
null	Process didn't start or was killed before exit

Always check exitCode rather than relying on stderr being empty. Many tools write to stderr even on success.

const result = await sandbox.exec("npm run lint");
if (result.exitCode !== 0) {
  console.error(`Lint failed:\n${result.stderr}`);
}

---

Concurrency Limits

A single sandbox can handle multiple concurrent executions. The API queues them internally and allows up to 3 concurrent executions per sandbox by default.

// These run concurrently inside the same sandbox
const [result1, result2, result3] = await Promise.all([
  sandbox.exec("echo task1 && sleep 2"),
  sandbox.exec("echo task2 && sleep 2"),
  sandbox.exec("echo task3 && sleep 2"),
]);

If you exceed the concurrency limit, requests return 429 Too Many Requests.

---

Common Errors

Status	Code	Description
400	INVALID_REQUEST	Missing command field
400	COMMAND_TOO_LONG	Command exceeds 64KB limit
404	SANDBOX_NOT_FOUND	Sandbox doesn't exist
408	EXECUTION_TIMEOUT	Command exceeded timeoutMs
409	INVALID_STATE	Sandbox is not in running state
429	RATE_LIMITED	Too many concurrent executions
502	EXECUTION_ERROR	Internal sandbox error

Next Steps

• Sandboxes API — manage sandbox lifecycle
• Running Code Guide — execution patterns and best practices
• CLI Reference — run commands via CLI