About The Agentic Leaderboard

The Agentic Leaderboard is a weekly-updated, fully automated benchmark of the top 100 open-source AI agents. It was launched in 2025 by Craig Wallace to provide a transparent, reproducible ranking that the community can rely on for discovering and comparing agentic AI tools.

Data Sources

• GitHub Search API — repos tagged with agentic keywords (stars ≥ 50)
• GitHub Fresh Search — repos created in the last 7 days (stars ≥ 5)
• Hugging Face Spaces & Models API
• Papers With Code API — research-backed agent projects
• Curated awesome-lists (e2b-dev/awesome-ai-agents, kyrolabs/awesome-agents)
• Manual seed list of known agentic products

Update Cadence

The pipeline runs every Monday at 11:00 UTC (7am ET) via GitHub Actions. Six agents execute in sequence: Scraper → Evaluator → Ranker → Publisher → Blaster → Broadcaster. The entire process takes ~30 minutes.

Scoring Methodology

Each agent receives an overall score from 0–100, calculated as:

• 35% Reliability success rate
• 20% Tool selection quality
• 15% Autonomous iteration efficiency
• 10% Cost + latency efficiency
• 20% Community mindshare (stars, forks, HF likes, paper citations)

Full details: Methodology page

Contact

For corrections, press inquiries, or partnership:info@theagenticleaderboard.com