shout-run-sdk

Python SDK for shout.run. Broadcast your terminal from code.

Write a script, pipe some output, and anyone with the link watches it live.

Install

pip install shout-run-sdk

Quick Start

from shout_sdk import ShoutSession

session = ShoutSession(api_key='shout_sk_...')
info = session.start(title='My Agent Session')
print(f"Live at: {info['url']}")

session.write('Hello, world!\r\n')
# ... do work ...

session.end()

Context Manager

from shout_sdk import ShoutSession

with ShoutSession(api_key='shout_sk_...') as session:
    info = session.start(title='My Agent')
    session.write('Working...\r\n')
    # session.end() called automatically

Getting an API Key

npm install -g shout-run
shout login
shout api-key create "My Agent"

The key is printed once. Save it somewhere safe. Keys start with shout_sk_.

You can list your keys with shout api-key list and revoke one with shout api-key revoke <id>.

API Reference

ShoutSession(api_key, **kwargs)

Methods

• start() — Creates the session and connects WebSocket. Returns {'session_id', 'url', 'ws_url'}.
• write(data) — Sends terminal output (str or bytes). Automatically buffered, rate-limited, and chunked.
• resize(cols, rows) — Updates terminal dimensions.
• end() — Flushes buffer, sends end frame, closes session.

Static Methods

• ShoutSession.delete_session(api_key, session_id, *, api_url='https://api.shout.run') — Delete an ended session (permanent). Raises RuntimeError on failure.

ShoutSession.delete_session(api_key='shout_sk_...', session_id='abc123')

• ShoutSession.search_sessions(api_key, query, *, tags=None, status=None, limit=20, cursor=None, api_url='https://api.shout.run') — Search sessions by query, tags, and status.

results = ShoutSession.search_sessions(
    api_key='shout_sk_...',
    query='typescript',
    tags=['tutorial'],
    status='ended',
    limit=10,
)
for session in results:
    print(f"{session['title']} by {session['username']}")

• ShoutSession.get_session_content(api_key, session_id, *, api_url='https://api.shout.run') — Get session metadata and plain-text transcript.

content = ShoutSession.get_session_content(api_key='shout_sk_...', session_id='abc123')
print(f"Title: {content['session']['title']}")
print(f"Transcript: {content['transcript'][:200]}...")

Properties

• state — Current state: SessionState.IDLE, .CONNECTING, .LIVE, .ENDING, .ENDED
• viewers — Current viewer count
• session_id — Session ID (None before start)

Events

session.on('connected', lambda: print('Connected!'))
session.on('disconnected', lambda code, reason: print(f'Disconnected: {code}'))
session.on('reconnecting', lambda attempt: print(f'Reconnecting #{attempt}'))
session.on('viewers', lambda count: print(f'Viewers: {count}'))
session.on('error', lambda err: print(f'Error: {err}'))
session.on('state_change', lambda state: print(f'State: {state}'))

Piping Subprocess Output

import subprocess
from shout_sdk import ShoutSession

session = ShoutSession(api_key='shout_sk_...')
info = session.start(title='Running tests')

proc = subprocess.Popen(
    ['pytest', '-v'],
    stdout=subprocess.PIPE,
    stderr=subprocess.STDOUT,
)

for line in proc.stdout:
    session.write(line)

proc.wait()
session.end()

License

MIT