Automation Agents

The system operates on a scheduled content pipeline using APScheduler. Each day at a configured time, it reads recently posted Instagram content and expands it into full-length blog articles (800–1500 words) via the Grok AI language model (grok-3) with a specialized brand-voice system prompt. A weekly category rotation cycles through product features, industry tips, education, data security, sustainability, and company news to maintain topical diversity. After generation, the Internal Linker engine scans article content and inserts contextual hyperlinks to relevant tool pages based on keyword matching. The SEO Analyzer then scores each draft (0–100) on title length, meta description, focus keyword density, word count, internal link count, featured image presence, and heading structure. Drafts enter a review queue where admins can edit, approve, or reject before automated publishing. Featured images are generated via Grok AI's image model and served from a dedicated output directory. Email notifications alert editors when new drafts are ready for review.

The system operates on a scheduled content calendar using APScheduler with configurable generation and posting times. Five content generators rotate through product features, industry tips, engagement questions, education posts, and site update announcements. Each generator produces a caption via the Grok AI language model with a brand-voice system prompt, and a 1080x1080 branded image via Pillow templates with customizable colors, fonts, and layouts. A site monitor checks the customer's website on a configurable interval for changes, triggering priority content when updates are detected. The hashtag engine rotates across 7 categories to maximize reach without repetition. Generated drafts enter a review queue where admins can edit, approve, or reject before automated posting via the Instagram Graph API. Email notifications alert editors when new drafts are ready for review.

The Sitemap Agent operates on a dual-schedule system using APScheduler. A primary daily job runs at 3 AM Pacific time to perform a complete sitemap regeneration, ensuring all pages are accounted for with fresh timestamps. A secondary job runs every 30 minutes specifically to monitor the blog section for newly published or removed posts. When blog changes are detected, the agent triggers an immediate sitemap update and submits the new URLs to search engines via the IndexNow protocol for near-instant indexing. The agent maintains a tracking database to detect changes efficiently without scanning the entire site. Each URL in the sitemap receives an appropriate priority score (1.0 for homepage, 0.9 for service pages, 0.7 for blog posts) and change frequency hint based on content type. All generation runs, detected changes, and search engine submissions are logged for monitoring and debugging.

The Code Review Agent operates on a scheduled pipeline using APScheduler, executing nightly at 2 AM Pacific time. Each scan traverses the /var/www directory structure, analyzing Python files across all registered services. The analysis engine runs multiple detection modules in sequence: Security Analysis detects SQL injection patterns, XSS vulnerabilities, command injection risks, hardcoded credentials, and insecure cryptographic practices. Dependency Analysis parses all requirements.txt files and compares installed versions against the PyPI registry to identify outdated packages with known security patches. Code Quality Analysis flags bare except clauses, debug mode enabled in production, print statements in web applications, and deprecated API usage. Best Practice Analysis checks for proper error handling, logging configuration, and Flask/SQLAlchemy patterns. Each finding is stored with file path, line number, code snippet, severity level, and remediation guidance. An admin dashboard provides filtering, bulk status updates, and manual review triggering. Critical and high-severity findings trigger immediate email alerts to the development team.