Downstream Provider Notifier

Automate the provision of information (Article 13) to downstream deployers when an AI system is updated.

When to use

Trigger when:

•"Notify our downstream providers about the latest code changes"
•"Send the Article 13 compliance package to deployers"
•"Summarize what changed in the system for our partners"

Workflow

code

Downstream Notification Progress:
- [ ] Step 1: Detect changes (Git Diff or SQLite Snapshot)
- [ ] Step 2: Generate AI summary using Gemini (Article 13.3.c focus)
- [ ] Step 3: Identify recipients from examples/ directory
- [ ] Step 4: Generate compliance packages (JSON Manifest + Excel + PDF)
- [ ] Step 5: Log transaction

Modes

•Git Mode: Automatic git diff detection for repositories.
•Snapshot Mode: SQLite-based file hashing (examples/notices.db) for tracking changes in local directories without git.

Usage

1. Preparation

Place master templates in the templates/ folder:

•templates/Article_13_Compliance_Template.xlsx
•templates/Annex_XII_Provider_Template.xlsx

Add provider-specific info as *_Provider.xlsx files in the examples/ directory.

2. Execution

Run the notifier against your codebase:

powershell

# Against a git repo
python scripts/notifier.py --repo "path/to/repo"

# Against a non-git directory (SQLite diffing)
python scripts/notifier.py --repo "path/to/folder"

3. Output

Each run creates a batch in Output/Batch_[TIMESTAMP]/:

•Notice_[ID].json: SMTP-ready manifest with To, Subject, and Attachment paths.
•Article_13_[ID].xlsx: Populated compliance template.
•Summary_[ID].pdf: AI-generated summary PDF.

Testing

Unit tests are in examples/tests/. Run them with:

powershell

python -m pytest examples/tests/test_notifier.py

Quality Checklist

• Code changes correctly identified
• AI Summary is accurate and non-technical
• All 3+ providers from registry are processed
• Article 13 template is correctly attached
• Audit log is updated with timestamps and recipient IDs