Executive Summary
Technical blogging and market analysis require consistent manual research, RSS feed parsing, drafting, and publishing, which often fall behind due to high client project workloads.
Stanley's Log automates the entire editorial lifecycle. A daily GitHub Action runs a headless scraper across 15+ tech news sources (TechCabal, Techpoint, Benjamindada, WeeTracker), filters political noise, synthesizes field notes using Gemini AI, and commits Markdown files straight to Git repository history.
Core Capabilities & System Highlights
15+ Tech Source Scraper & Filter
Parses RSS feeds and web articles across top African tech & startup publishers, filtering out geopolitical noise to focus strictly on technology & venture insights.
Multi-Model Resiliency Fallback
Uses Gemini 2.5 Flash as primary LLM generator with automatic fallback to Gemini 2.5 Pro if rate limits or API throttles occur.
GitOps Markdown Persistence
Articles are committed directly to `/content/blog` as version-controlled Markdown files, leaving an immutable audit trail with zero database hosting costs.
History Deduplication Engine
Tracks processed subjects inside `agent-history.json` to prevent repetitive content generation across daily publishing runs.
System Architecture & Specs
Gemini 2.5 API & Multi-Model Chain
Prompt-engineered LLM pipeline producing structured JSON containing markdown body, category tags, read times, and summaries.
Cheerio & RSS Parser Pipeline
Fetches live RSS feeds and web pages, stripping HTML boilerplate and extracting core article text for AI prompt context.
Scheduled GitHub Actions Workflow
Executes CLI script (`scripts/agent.ts`) daily via GitHub Actions, committing new Markdown files directly to the main branch.
Key Engineering Trade-offs & Operational Notes
GitHub Actions trigger research runs on a timer without keeping a server awake.
Articles are saved directly to the code repository, ensuring full revision control.
Automatic fallback to secondary AI models prevents failed runs when an API rate limit is reached.
Tech Stack & Role Matrix
Every library and framework in Stanley’s Log was chosen with intentional architectural trade-offs to balance speed, type safety, security, and scalability.
| Technology | Category | System Role | Performance Rationale |
|---|---|---|---|
| Next.js | Framework | Blog Rendering & MDX Content Pipeline | Statically generates blog pages for zero-latency reading. |
| Gemini AI | AI / ML | Autonomous Research & Synthesis | Delivers deep contextual understanding and structured markdown generation. |
| GitHub Actions | CI/CD | Headless Daily Cron Execution | Eliminates server hosting costs by running on-demand scheduled jobs. |
| TypeScript | Language | CLI Script & Pipeline Logic | Prevents runtime failures during news parsing and file writing. |
| Cheerio | Scraping | Fast HTML Parsing & Content Extraction | Lightweight server-side HTML scraping without heavy browser overhead. |
Live Preview & Showcase
Metrics & Reliability
Publishing Cost
Runs via free-tier GitHub Actions & Gemini API
Pipeline Reliability
Multi-model AI fallback prevents failed runs
Content Control
Articles are version-controlled Markdown files