paldom/llm-council-skills

Decide

Whether a council is worth it at all - evidence-based triage before you pay the multi-model multiplier.

Design

The pipeline, the members, and the prompts - fan-out, anonymized peer review, and synthesis that preserves disagreement.

Operate

Keep it correct and affordable in production - robust aggregation, cost gating, and failure-mode defenses.

Implement

Run real coding harnesses as council members from plain Python.

Govern

The human oversight layer - councils of people, not models.

Other skills