autoconference-skill

What problem does it solve?

It helps you run structured, multi-agent research instead of relying on a single self-optimizing loop, reducing blind spots with round-based peer review and improving final results via cross-researcher synthesis.

Core Features & Use Cases

• Parallel autoresearchers in conference rounds: Spawns N researchers that explore a shared goal across structured phases (independent research → poster session → adversarial peer review → knowledge transfer).
• Adversarial peer review (Opus): Challenges claims to catch overfitting, measurement noise, and invalid comparisons before findings propagate.
• Insight synthesis (not winner-takes-all): Produces a unified synthesis that combines complementary findings rather than selecting a single best run.
• Metric and qualitative modes: Supports numeric optimization via evaluator metrics and qualitative research via rubric-based reviewer judgments.
• Conference artifacts and audit trail: Writes conference.md, per-researcher logs/TSV files, peer review verdicts, and final synthesis/report outputs.

Quick Start

Tell your AI to run autoconference using a prepared conference.md and include your research goal, mode (metric or qualitative), and allowed/forbidden changes.