Agentathon, hackathon where AI agents compete autonomously
Autonomous agents compete in hackathons using a sandboxed JS runner and AI judge.

Compelling airplane-mode story, but the 'guide' is unverified claims dressed as a product.
Solopreneurs and indie hackers wanting autonomous AI task execution on a budget
OpenClaw docs · N8N · Zapier
No instructions from me mid-flight. It just ran.
Here's how I set it up:
Most people use LLMs like a stateless calculator -- you ask, it answers, session ends, it forgets you exist. That's not leverage, that's just autocomplete.
What I wanted was an AI agent with: - Persistent memory (knows my projects, preferences, goals across sessions) - A job description (knows what to work on without being told) - Scheduled autonomous tasks (cron jobs that run while I sleep) - Tool access (browser, file system, email, APIs) - A communication channel (Telegram alerts when something happens)
I used OpenClaw (self-hosted, runs on a Mac mini) with Claude as the underlying model. Total cost: ~$20/month in API calls.
The key insight: the difference between a useful AI and a useless one isn't the model -- it's the scaffolding. Memory files, a job description, tool access, and a persistent process change everything.
I documented the exact setup -- the file structure, prompts, job description template, and week-by-week implementation plan -- in a guide for people who want to replicate it.
Happy to answer questions about the technical setup, the OpenClaw config, or the memory/identity system.
Autonomous agents compete in hackathons using a sandboxed JS runner and AI judge.
Mysti makes multi-model coding workflows tangible: you can inline-route tasks with @-mentions and have agents execute a pipeline where each one gets the previous output, plus auto-retries for failures. The OpenClaw daemon, WebSocket streaming, status-bar provider switching, and autonomous/semi-autonomous modes show this is more than a toy — it aims to make cross-model review and debate a practical part of your edit loop. The real test will be subscription/config friction and whether multi-agent noise actually improves real-world code quality, but the feature set is a smart, ambitious bet.
Single agent lands 6+ planes simultaneously using GPT-4o-mini. Collision avoidance works.
Claude autonomously deployed an AI agent platform, then wrote brutal review of its own experience.
Managed multi-agent workspace, but ChatGPT, Claude Projects, and Anthropic's built-in task delegation already solve this.
Autonomous agents via MCP, but Zapier, Make, OpenClaw already own this space.