What problem does it solve?
Standard prompt generation relies on self-critique, which has inherent structural blind spots that often miss the exact flaws present in the original prompt. This is a minor issue for throwaway prompts, but for high-stakes use cases like production content moderation, compliance checks, or legal escalation workflows, these missed flaws can lead to significant harm, financial loss, or regulatory penalties.
Core Features & Use Cases
- Validated Prompt Generation Pipeline: Runs 5 sequential validation circles (C1 STOP, C2 RESEARCH, C3 GRID, C4 TRIBUNAL, C5 FIX) with structured JSON traces, plus an 18-hacks filter to optimize prompts for tokens, speed, security, or quality based on your specific needs.
- Optional 5-Advisor Council Audit: For high-risk prompts, activate a multi-agent council with 5 distinct advisors (Contrarian, First Principles Thinker, Expansionist, Outsider, Executor) that perform blind peer review to detect flaws missed by self-critique, plus a Chairman synthesis with a clear verdict and one concrete immediate action.
- Compliance & Safety Guardrails: Automatically flags correlated proxy variables (e.g., zip code carrying forbidden origin signal) for compliance/legal domains, and mandates full human escalation workflow details (who processes, timeline, context, decision recording) for prompts requiring manual review.
- Audit Artifacts: When the council is activated, generates a visual HTML report and full markdown transcript of the entire deliberation process for documentation, auditing, and future reference.
- Use Case Example: Create a prompt for moderating user-generated content in a production system, then run a council audit to catch edge cases like sarcasm, dogwhistles, and ambiguous "borderline content" definitions that a self-critique would completely miss.
Quick Start
Use the promptor-council skill to create a production-ready content moderation prompt and request a council audit to validate it for edge cases and ambiguous scenarios.
