Autoresearch

This is a full autonomy research orchestrator that runs the entire pipeline from literature review to paper publication without asking permission. It maintains a two-loop experiment and synthesis cycle, routes work to domain-specific skills, and updates you through periodic HTML or PDF reports while you're away. The core idea is that it acts as project manager, not domain expert. You either bootstrap from a vague idea, hand it a clear question, or resume from existing state. Security audits show some warnings, so review what it's doing before letting it loose on anything sensitive. It's genuinely hands-off if you trust the approach, but that autonomy cuts both ways.

Install to Claude Code

npx -y skills add orchestra-research/ai-research-skills --skill autoresearch --agent claude-code