title	Synaptrix MailOS OpenEnv
emoji	📧
colorFrom	blue
colorTo	green
sdk	docker
app_port	7860

Synaptrix MailOS OpenEnv

Synaptrix MailOS is an AI-powered multi-agent email decision intelligence system that classifies emails, detects urgency and risk, automates Gmail-style routing, generates tasks and reply drafts, and exposes an OpenEnv-compatible benchmark with graded tasks and reward-based evaluation.

Features

• Multi-agent email understanding
  • classifies emails as Work, Personal, or Spam
• Risk and urgency analysis
  • estimates priority, tone, spam risk, and threat indicators
• Decision intelligence engine
  • recommends Ignore, Respond, or Urgent Action
• Gmail automation simulation
  • applies labels, routes folders, and organizes emails into smart inbox lanes
• Smart inbox organizer
  • separates emails into Urgent, Important, and Others
• Auto task generation
  • creates tasks from deadlines, meetings, and time-sensitive messages
• Auto-reply system
  • generates context-aware reply drafts and supports simulated sending
• Analytics and history
  • shows processed emails, rewards, risk trends, and action history
• OpenEnv benchmark support
  • includes typed models, reset(), step(), state(), 3 benchmark tasks, graders, and baseline scoring

How To Run

1. Install dependencies:

pip install -r requirements.txt

2. Start the app:

python app.py

3. Open the project in your browser:

http://127.0.0.1:8000

How To Use

1. Open the website.
2. Click Analyze Email to let the AI study the current email.
3. Check the result in Summary, Automation, Tasks, Reply, and Analytics.
4. Use manual actions like Ignore, Respond, or Urgent Action if you want to override the AI.
5. Open Composer to test your own custom email.
6. Open Settings to change theme, color vibe, density, and behavior.

OpenEnv Endpoints

These routes are available for the benchmark environment:

• GET /health
• GET /tasks
• POST /reset
• POST /step
• GET /state
• GET /metadata
• GET /schema

The UI uses these routes:

• GET /api/state
• POST /api/reset
• POST /api/analyze
• POST /api/step
• POST /api/open-email
• POST /api/send-reply

Benchmark Tasks

The project includes 3 benchmark tasks:

• easy-spam-ignore
  • detect obvious spam and ignore it safely
• medium-reply-scheduling
  • recognize a valid personal email and respond correctly
• hard-security-escalation
  • identify a critical work/security issue and escalate it urgently

Baseline

Run the baseline scorer with:

python baseline.py

If OPENAI_API_KEY is available, the script can use the OpenAI client. Otherwise, it falls back to a deterministic reference policy and still produces reproducible scores in baseline_scores.json.

Project Structure

• email_intelligence/
  • backend logic, agents, environment, tasks, graders, and API
• static/
  • frontend pages, styles, and scripts
• app.py
  • main application entrypoint
• baseline.py
  • baseline evaluation runner
• openenv.yaml
  • environment metadata
• Dockerfile
  • container setup for deployment

Deployment

This project includes:

• Dockerfile
• requirements.txt
• openenv.yaml

So it is ready to be containerized and used for Hugging Face Space style deployment.