P1 PG Production Flow

Generates DATS section narratives through parallel threads, convergence audits, and quality gating.

Updated Apr 11, 2026
One-click install
npx skills add https://github.com/gozonerd/shadow-ai-assessment --skill p1-pg-production-flow-gozonerd
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: P1 PG Production Flow
Source: https://github.com/gozonerd/shadow-ai-assessment/tree/main/.claude/skills/dmis/pipeline/P1_PG_Production_Flow
Command: npx skills add https://github.com/gozonerd/shadow-ai-assessment --skill p1-pg-production-flow-gozonerd

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Producing long-form, compliance-grade narrative sections requires consistency, source fidelity, and verifiable quality, which single-pass generation cannot guarantee. This pipeline coordinates multi-thread generation, layered auditing, and gated synthesis to produce validated section content. ## Core Features & Use Cases - Parallel Thread Generation: Runs 15 parallel generation threads to produce candidate section outputs for comparative ranking. - 8-Layer Convergence Audit: Verifies requirements compliance, regulatory cross-references, coverage, source integrity, and contradictions with a Krippendorff alpha threshold of 0.70. - Synthesis with Quality Gate: Merges the winning thread into a final section, enforces 4 post-synthesis checks and a 10 KB minimum size, and halts or remediates on failure. - Use Case: A documentation team generating a regulatory DATS section runs the pipeline to obtain a gated, audit-passed narrative with Telegram notifications at each milestone. ## Quick Start Run the P1 PG Production Flow pipeline to generate and quality-gate the next DATS section narrative.

Frequently Asked Questions about P1 PG Production Flow

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How does the P1 PG Production Flow generate section content?

It runs 15 parallel generation threads to produce candidate outputs, audits them across 8 convergence layers, synthesizes the winning thread, and applies a final quality gate before marking the section complete.

What is the convergence threshold for the audit layers?

The pipeline requires a Krippendorff alpha of at least 0.70 across the 8 audit layers. Each layer gets one critique-rewrite retry cycle before the convergence judge decides between synthesis, halt, or remediation.

What happens when a post-synthesis check fails?

The flow returns to the synthesis agent with a refined prompt, allowing a maximum of 2 retry cycles. If the quality gate still fails, the issue escalates to a human decision on rework versus section redesign.

Why does the quality gate reject outputs under 10 KB?

P1 sections are narrative-heavy, so outputs below 10240 bytes indicate content gaps. The gate fails automatically on undersized files regardless of other passing criteria.

How long does a full pipeline run take per section?

The estimated total is 24 to 37 minutes per section: 4-8 minutes for thread generation, 12-16 minutes for the convergence audit, 6-10 minutes for synthesis and checks, and 2-3 minutes for the gate decision.