Enhance deployment and CI/CD workflows with API improvements and integrations

[stenwire]

No description provided.

[gitguardian]

⚠️ GitGuardian has uncovered 6 secrets following the scan of your pull request.

Please consider investigating the findings and remediating the incidents. Failure to do so may lead to compromising the associated services or software components.

🔎 Detected hardcoded secrets in your pull request

GitGuardian id	GitGuardian status	Secret	Commit	Filename
29770111	Triggered	Generic Password	ec5775f	frontend/tests/unit/api.test.ts	View secret
29770111	Triggered	Generic Password	ec5775f	frontend/tests/unit/api.test.ts	View secret
29802212	Triggered	Generic CLI Secret	d6f49d3	backend/scripts/create_admin.py	View secret
29802212	Triggered	Generic CLI Secret	0ad9370	backend/scripts/create_admin.py	View secret
29802212	Triggered	Generic CLI Secret	ac68983	backend/scripts/create_admin.py	View secret
29802212	Triggered	Generic CLI Secret	1c0a69f	backend/scripts/create_admin.py	View secret

🛠 Guidelines to remediate hardcoded secrets

1. Understand the implications of revoking this secret by investigating where it is used in your code.
2. Replace and store your secrets safely. Learn here the best practices.
3. Revoke and rotate these secrets.
4. If possible, rewrite git history. Rewriting git history is not a trivial act. You might completely break other contributing developers' workflow and you risk accidentally deleting legitimate data.

To avoid such incidents in the future consider

• following these best practices for managing and storing secrets including API keys and other credentials
• install secret detection on pre-commit to catch secret before it leaves your machine and ease remediation.

---

^{🦉 GitGuardian detects secrets in your source code to help developers and security teams secure the modern development process. You are seeing this because you or someone else with access to this repository has authorized GitGuardian to scan your pull request.}

[github-actions]

🤖 Hi @stenwire, I've received your request, and I'm working on it now! You can track my progress in the logs for more details.

[github-actions]

📋 Review Summary

This Pull Request introduces significant new features including a multi-agent AI system, comprehensive WhatsApp integration for campaigns and messaging, a full-fledged subscription management system with Paystack, and an administrative panel for data management. The changes also include substantial refactoring of core components for better configuration, security, and maintainability. The addition of detailed documentation for contributors and project context is a significant improvement.

🔍 General Feedback

• The architectural shift to a multi-agent system and the integration of external services (WhatsApp, Paystack) are well-structured, leveraging factory patterns and clear abstractions.
• Security considerations, especially in prompt injection defense and response sanitization, are robust and well-implemented.
• Extensive new Alembic migrations and API endpoints demonstrate a significant expansion of the application's capabilities.
• Configuration management has been greatly improved with environment-specific settings.
• Several minor inconsistencies and potential areas for further robustness and efficiency have been noted in inline comments, especially regarding database session management in certain async paths and redundant file/code. Some logging could also be refined for production use. Overall, a high-quality and impactful set of changes.

[github-actions]

🟡 This file appears to be a direct copy of deploy.yml. It's redundant and should be removed.

Suggested change

	# Remove this file as it's a redundant backup of deploy.yml

[github-actions]

🟢 The removal of .env from backend/.gitignore is fine because the root .gitignore now handles .env*. However, backend/.gitignore still includes __pycache__/ and other build artifacts which are already covered by the root .dockerignore and .gitignore. It would be cleaner to keep backend/.gitignore minimal if the root one covers most common ignores.

[github-actions]

🟡 The utcnow function is defined in app.models.whatsapp_broadcast.py but is not consistently used across all models that define default=lambda: datetime.now(timezone.utc). It would be better to have a single, consistently imported utcnow function (e.g., from app.models.mixins or app.core.utils) and use it everywhere for uniformity and to avoid potential timezone issues.

[github-actions]

🟠 In get_analytics_overview, the leads_captured query uses GuestUser.is_lead.is_(True). However, the migration 35c884792568_add_is_lead_to_guest_users.py adds is_lead with nullable=True. This inconsistency might lead to unexpected behavior if is_lead can be None in the database. The query should handle None values if they are possible, or the column should be explicitly nullable=False in the migration, with a server default. The GuestUserResponse schema already accounts for None by setting it to False in model_validate, which indicates this is a known potential state.

[github-actions]

🟡 In _enrich_plan_fields, ast.literal_eval is used to parse the features field from the Plan model. If features is intended to be stored as a JSON string in the database, json.loads() would be more appropriate and explicit. If it's stored as JSONB, SQLAlchemy usually handles deserialization automatically, making this parsing redundant and potentially indicating an underlying data storage inconsistency for the JSON column type.

[github-actions]

🟡 The file backend/app/services/agent_system/agents.py still exists and defines agents like rag_agent. However, agent_factory.py now dynamically creates and configures these agents through methods like create_chief_agent, which is then used by agent_service.py. It appears agents.py might be an outdated artifact or serves a purpose not immediately clear. Consider removing or refactoring it if its content is no longer actively used by the main agent system or is fully superseded by agent_factory.py to avoid confusion and maintain a cleaner codebase.

[github-actions]

🟢 In _normalize, the regex r"[\s\-_.*|/\]+" is used to collapse repeated punctuation and whitespace. While functional, a minor optimization could be explored to potentially combine this with other text normalization steps (e.g., text.translate(_LEET_MAP)) if doing so improves efficiency without sacrificing readability or maintainability. This is a low-priority optimization.

[github-actions]

🟡 The email sending mechanism within escalate_to_human uses asyncio.new_event_loop() in a separate thread. While this prevents blocking the main FastAPI event loop, managing asyncio event loops in separate threads can be complex and introduce subtle issues. For more robust background task handling in a FastAPI application, consider integrating a dedicated task queue (e.g., Celery, FastAPI Background Tasks for short-lived tasks, or a more sophisticated approach for long-running ones) which is better suited for reliability and error handling in a production environment.

[github-actions]

🟡 Similar to the escalation email, the order notification email in create_order also uses asyncio.new_event_loop() within a separate thread. This approach, while non-blocking, carries the same considerations and potential complexities regarding asyncio event loop management in threads. A more robust solution for background email sending would involve a dedicated task queue for production readiness.

[github-actions]

🟡 The currency in the order notification email body is hardcoded based on the currency of the first item in the items list. It would be more accurate and robust to use order.currency after the Order object has been created and populated, ensuring consistency with the overall order details. Additionally, if an order can contain items with different currencies (though typically a single order has one currency), this logic should account for that.

[gemini-code-assist]

Code Review

This pull request introduces significant features to the TaimakoAI platform, including a robust Paystack-powered subscription billing system, a product catalogue with order placement capabilities, and a comprehensive WhatsApp broadcast campaign system supported by a standalone background worker. It also adds a secure administrative panel via sqladmin and enhances security and rate-limiting middleware. The review feedback highlights critical improvements for local development and production stability, specifically addressing SQLite compatibility issues with PostgreSQL-specific SQL functions, preventing potential AttributeError crashes on unvalidated database queries, improving the resilience of CSV imports against encoding and header variations, and securing configuration parsing against empty environment variables.

Important

The consumer version of Gemini Code Assist on GitHub is being sunset. Starting June 18, 2026, new organization installations will be blocked, and all code review activity will officially cease on July 17, 2026.
For more details on the timeline and next steps, please review the Help Documentation.

[gemini-code-assist]

SQLite (used for local development and testing) does not support the FOR UPDATE SKIP LOCKED syntax and will raise a syntax/operational error. Conditionally appending this clause only when running on PostgreSQL ensures the worker is fully functional and testable in local environments.

Suggested change

	now = datetime.now(timezone.utc)
	# Use raw SQL for FOR UPDATE SKIP LOCKED (portable-ish; PostgreSQL required in prod).
	result = db.execute(
	text(
	"""
	SELECT id FROM whatsapp_campaigns
	WHERE status = :scheduled
	AND scheduled_at <= :now
	ORDER BY scheduled_at ASC
	LIMIT :limit
	FOR UPDATE SKIP LOCKED
	"""
	),
	{
	"scheduled": CampaignStatus.SCHEDULED.value,
	"now": now,
	"limit": CLAIM_LIMIT,
	},
	)
	now = datetime.now(timezone.utc)
	dialect = db.bind.dialect.name
	for_update = "FOR UPDATE SKIP LOCKED" if dialect == "postgresql" else ""
	query = f"""
	SELECT id FROM whatsapp_campaigns
	WHERE status = :scheduled
	AND scheduled_at <= :now
	ORDER BY scheduled_at ASC
	LIMIT :limit
	{for_update}
	"""
	result = db.execute(
	text(query),
	{
	"scheduled": CampaignStatus.SCHEDULED.value,
	"now": now,
	"limit": CLAIM_LIMIT,
	},
	)

[gemini-code-assist]

Calling new_plan.to_dict() before checking if new_plan is None will raise an AttributeError if the plan is not found, leading to a 500 Internal Server Error instead of a 404 Not Found.

Suggested change

	print(f"\n\n New Plan: {new_plan.to_dict()}\n\n")
	if not new_plan:
	raise HTTPException(status_code=404, detail="New plan not found or inactive")
	if not new_plan:
	raise HTTPException(status_code=404, detail="New plan not found or inactive")
	print(f"\n\n New Plan: {new_plan.to_dict()}\n\n")

[gemini-code-assist]

If metadata is explicitly null in the JSON payload, event.get("raw_payload", {}).get("data", {}).get("metadata", {}) can return None. In that case, isinstance(metadata, str) is False, and metadata.get("is_upgrade", False) will raise an AttributeError. Using or {} ensures metadata is always a dictionary.

    metadata = event.get("raw_payload", {}).get("data", {}).get("metadata") or {}
    if isinstance(metadata, str):
        import json
        try:
            metadata = json.loads(metadata)
        except (ValueError, TypeError):
            metadata = {}

    is_upgrade = metadata.get("is_upgrade", False)

[gemini-code-assist]

If metadata is explicitly null in the JSON payload, event.get("raw_payload", {}).get("data", {}).get("metadata", {}) can return None. In that case, isinstance(metadata, str) is False, and metadata.get("is_upgrade", False) will raise an AttributeError. Using or {} ensures metadata is always a dictionary.

Suggested change

	metadata = event.get("raw_payload", {}).get("data", {}).get("metadata", {})
	if isinstance(metadata, str):
	import json
	try:
	metadata = json.loads(metadata)
	except (ValueError, TypeError):
	metadata = {}
	is_upgrade = metadata.get("is_upgrade", False)
	metadata = event.get("raw_payload", {}).get("data", {}).get("metadata") or {}
	if isinstance(metadata, str):
	import json
	try:
	metadata = json.loads(metadata)
	except (ValueError, TypeError):
	metadata = {}
	is_upgrade = metadata.get("is_upgrade", False)

[gemini-code-assist]

The now() function is PostgreSQL-specific and does not exist in SQLite, which will cause session updates to fail with an operational error during local development and testing. Using CURRENT_TIMESTAMP is standard SQL and is fully compatible with both PostgreSQL and SQLite.

Suggested change

	query = text("""
	UPDATE sessions
	SET state = :state, update_time = now()
	WHERE app_name = :app_name AND user_id = :user_id AND id = :session_id
	""")
	query = text("""
	UPDATE sessions
	SET state = :state, update_time = CURRENT_TIMESTAMP
	WHERE app_name = :app_name AND user_id = :user_id AND id = :session_id
	""")

[gemini-code-assist]

If these environment variables are defined as empty strings in the .env file, calling int() on them will raise a ValueError on startup. Using or with the default value handles empty strings gracefully.

Suggested change

	JWT_EXPIRATION_MINUTES: int = int(os.getenv("JWT_EXPIRATION_MINUTES", 60))
	REFRESH_TOKEN_EXPIRATION_DAYS: int = int(os.getenv("REFRESH_TOKEN_EXPIRATION_DAYS", 30))
	JWT_EXPIRATION_MINUTES: int = int(os.getenv("JWT_EXPIRATION_MINUTES") or 60)
	REFRESH_TOKEN_EXPIRATION_DAYS: int = int(os.getenv("REFRESH_TOKEN_EXPIRATION_DAYS") or 30)

[gemini-code-assist]

The new Google GenAI SDK (google-genai) is imported directly in multiple files (e.g., business.py, analysis_agent.py, tools.py), but it is not explicitly declared in pyproject.toml. Relying on transitive dependencies from google-adk is risky and can lead to broken environments if dependencies change upstream.

Suggested change

	"google-adk>=1.14.1",
	"google-generativeai>=0.8.5",
	"google-adk>=1.14.1",
	"google-genai>=0.1.1",
	"google-generativeai>=0.8.5",

[gemini-code-assist]

Using next(get_db()) to manually extract a database session from a FastAPI dependency generator is an anti-pattern. It leaves the generator suspended at the yield statement until garbage collection. Importing and using SessionLocal() directly is the standard and safer way to obtain a database session outside of request dependency injection.

        from app.db.session import SessionLocal
        db = SessionLocal()

[gemini-code-assist]

If SMTP_PORT is defined as an empty string in the .env file (e.g., SMTP_PORT=), os.getenv("SMTP_PORT", 587) will return "", and calling int("") will raise a ValueError on startup. Using or 587 handles empty strings gracefully.

Suggested change

	SMTP_PORT: int = int(os.getenv("SMTP_PORT", 587))
	SMTP_PORT: int = int(os.getenv("SMTP_PORT") or 587)

[gemini-code-assist]

Decoding the CSV content with 'utf-8' can leave a Byte Order Mark (BOM) like \ufeff at the beginning of the first header, causing key lookups (like row.get('sku')) to fail if the file was saved with a BOM (e.g., from Excel). Using 'utf-8-sig' handles this gracefully.

Suggested change

	content = await file.read()
	decoded = content.decode('utf-8')
	content = await file.read()
	decoded = content.decode('utf-8-sig')

[github-actions]

## 📋 Security Analysis Summary

I have analyzed the changes in this pull request and have not found any new security vulnerabilities. The changes appear to be safe from a security perspective.

🔍 General Feedback

• The pull request focuses on improving deployment and CI/CD workflows, with no changes to the application logic.
• The changes are well-structured and follow best practices for infrastructure as code.