Job submission and approval workflow for SyftBox (execution handled by syft-simple-runner)
A simple, lightweight system for executing code on remote SyftBox datasites with manual approval workflows.
Syft Code Queue provides a clean separation between data scientists who submit code for execution and data owners who review and approve that code. All code execution requires explicit manual approval - there is no automatic approval built into the core system.
Data Scientist → Submit Code → Data Owner Reviews → Manual Approve → Execute → Results
run.sh scriptsimport syft_code_queue as q
# Submit jobs to others
job = q.submit_job(
target_email="data-owner@university.edu",
code_folder="./my_analysis",
name="Statistical Analysis",
description="Aggregate statistics computation",
tags=["statistics", "privacy-safe"]
)
# Or submit a simple script
job = q.submit_script(
target_email="data-owner@university.edu",
script_content="print('Hello, world!')",
name="Hello World Test"
)
# Monitor your jobs
q.my_jobs() # Jobs you've submitted
q.pending_for_me() # Jobs submitted to you for approval
q.approve("job-id", "Looks safe") # Approve a job
q.status() # Overall status
print(f"Job submitted: {job.uid}")
print(f"Status: {job.status}") # Will be 'pending'
In Jupyter notebooks, jobs display as beautiful interactive cards:
# Individual job with interactive buttons
job # Shows approval/rejection buttons you can click
# Job collections with filterable tables
q.jobs_for_me # Interactive table with batch approve/reject
q.jobs_for_others # Table showing your submitted jobs with logs/output buttons
q.pending_for_me # Shows only jobs awaiting your approval
Interactive Features:
import syft_code_queue as q
# View jobs submitted to you
q.pending_for_me() # Jobs waiting for your approval
q.all_jobs_for_me() # All jobs submitted to you
# Review and approve/reject jobs
q.review_job("job-id") # Get job details
q.approve("job-id", "Looks safe") # Approve job
q.reject("job-id", "Too broad") # Reject job
# Overall status
q.status() # Your queue status
# CLI Tools for Job Management (optional)
scq pending # List jobs pending approval
scq review a1b2c3d4 # Review specific job details
scq approve a1b2c3d4 -r "Looks safe" # Approve job
scq reject a1b2c3d4 -r "Too broad" # Reject job
scq list # List all jobs
scq status # Show queue status
pip install syft-code-queue
We provide role-specific tutorials for different users:
examples/Part 1 — Data Scientist Tutorial.ipynb - Learn to submit and monitor jobsexamples/Part 2 — Data Owner Tutorial.ipynb - Learn to review and approve jobsThe core design principle is manual approval only:
Any automation must be external and call the manual approval CLI or API:
# External automation example using CLI
import subprocess
def smart_approval_bot():
# Get pending jobs
result = subprocess.run(['scq', 'pending'], capture_output=True, text=True)
# Parse and approve based on criteria
for job_id in get_job_ids_from_output(result.stdout):
if meets_my_criteria(job_id):
subprocess.run(['scq', 'approve', job_id, '-r', 'Auto-approved by bot'])
else:
subprocess.run(['scq', 'reject', job_id, '-r', 'Does not meet criteria'])
See the examples directory for more automation examples.
Every job submission must be a folder containing:
my_analysis/
├── run.sh # Main execution script (required)
├── analyze.py # Your analysis code
├── requirements.txt # Python dependencies (optional)
└── README.md # Documentation (optional)
run.sh:#!/bin/bash
set -e
echo "Starting analysis..."
# Install dependencies
if [ -f requirements.txt ]; then
pip install -r requirements.txt
fi
# Run analysis
python analyze.py
echo "Analysis complete!"
SafeCodeRunner with timeouts and resource limits📤 submit → ⏳ pending → ✅ approved → 🏃 running → 🎉 completed
↘ 🚫 rejected ↘ ❌ failed
aggregate-analysis, statisticsq.review_job() to examine job details before approvingimport syft_code_queue as q
# Submit jobs
job = q.submit_job(target_email, code_folder, name, description, tags)
job = q.submit_script(target_email, script_content, name, description, requirements, tags)
# Monitor your submitted jobs
q.my_jobs() # All your jobs
q.get_job(job_uid) # Specific job
q.get_job_output(job_uid) # Job output
q.get_job_logs(job_uid) # Job logs
q.wait_for_completion(job_uid) # Wait for completion
# Manage jobs submitted to you
q.pending_for_me() # Jobs waiting for approval
q.all_jobs_for_me() # All jobs submitted to you
q.review_job(job_uid) # Review job details
q.approve(job_uid, reason) # Approve job
q.reject(job_uid, reason) # Reject job
# Status and help
q.status() # Overall status
q.help() # Show help
# List pending jobs
scq pending
# Review job details
scq review <job_id>
# Approve/reject jobs
scq approve <job_id> --reason "Approved because..."
scq reject <job_id> --reason "Rejected because..."
# Monitor jobs
scq list # All jobs
scq status # Queue status
scq --help # All available commands
To enable syft-code-queue on your datasite:
syft-code-queue folder to your datasiterun.sh to process jobsscq commands to manage job approvalsThe app uses sensible defaults, but can be customized:
from syft_code_queue import QueueConfig
config = QueueConfig(
queue_name="code-queue",
max_concurrent_jobs=3,
job_timeout=600, # 10 minutes
cleanup_completed_after=86400 # 24 hours
)
The run.sh script handles the entire queue processing cycle:
git clone <repository>
cd syft-code-queue
# Install in development mode
pip install -e .
# Run tests
pytest
# Run examples
python examples/example_cross_datasite_workflow.py
See CONTRIBUTING.md for development guidelines.
Licensed under the Apache License 2.0. See LICENSE file for details.
Simple. Secure. Manual. 🚀
FAQs
Job submission and approval workflow for SyftBox (execution handled by syft-simple-runner)
We found that syft-code-queue demonstrated a healthy version release cadence and project activity because the last version was released less than a year ago. It has 1 open source maintainer collaborating on the project.
Did you know?
Socket for GitHub automatically highlights issues in each pull request and monitors the health of all your open source dependencies. Discover the contents of your packages and block harmful activity before you install or update your dependencies.
Application Security
/Security News
Socket CEO Feross Aboukhadijeh and a16z partner Joel de la Garza discuss vibe coding, AI-driven software development, and how the rise of LLMs, despite their risks, still points toward a more secure and innovative future.
Research
/Security News
Threat actors hijacked Toptal’s GitHub org, publishing npm packages with malicious payloads that steal tokens and attempt to wipe victim systems.
Research
/Security News
Socket researchers investigate 4 malicious npm and PyPI packages with 56,000+ downloads that install surveillance malware.