ai-agent-knowledge

From trending noise to curated intelligence.

🤖 Maintenance

This repository is maintained autonomously by AI agents running on my VPS (cron-driven, daily). All architecture, scoring weights, source selection, dedup logic, and operational infrastructure were designed and deployed by me. The agents handle the day-to-day ingestion; I own the system design and review the output.

If you're evaluating this work: the engineering is in how the ingestion pipeline works (cron scheduling, scoring formulas, dedup memory, daily/weekly/monthly aggregation), not in the raw counts.

---

Autonomous daily ingestion system that discovers, scores, and documents AI tools relevant to Rajesh's stack.

Structure

skills/
├── coding-agents/     — Claude Code, OpenCode, Codex, Hermes, OpenClaw, Copilot, Devin
├── app-builders/      — Bolt.new, v0, Lovable, Replit Agent, Cursor
├── frameworks/        — LangGraph, AutoGen, CrewAI, AntiGravity, Mastra
├── infrastructure/    — Supabase, Redis, Ollama, OpenRouter, Vercel
├── model-providers/   — OpenAI, Anthropic, Google, xAI, Mistral
daily/                 — Daily activity logs (timestamped)
weekly/                — Weekly summaries and trend analysis
monthly/               — Monthly ecosystem snapshots
stats/                 — Machine-readable trend data (JSON)

Scoring System

Score = stars * 0.35
      + weeklyGrowth * 0.30
      + ecosystemMentions * 0.15
      + repoAgeFactor * 0.10
      + personalRelevance * 0.10

Where personalRelevance is based on Rajesh's stack:
Next.js · TypeScript · AI/LLMs · Supabase · Vercel · Python · React

How It Works

1. Observe — Scan GitHub trending across AI agent categories
2. Score — Rank by formula + personal stack relevance
3. Curate — Star top-scored repos (not just trending)
4. Learn — Write SKILL.md docs for new tools discovered
5. Store — Commit daily logs, weekly trends, monthly snapshots
6. Report — Deliver summary to Telegram

Built by RajeshKalidandi · Powered by Hermes Agent