# Pipeline Guard

> Validates and optimizes run_pipeline DOT graphs with model selection from QoS catalog

- Skill: `zhanghandong/pipeline-guard` (Agent Skill, multi-file: 4 files)
- Install (CLI): `npx skillmds@latest add zhanghandong/pipeline-guard`
- Raw SKILL.md: https://api.skillmd.com/api/skills/zhanghandong/pipeline-guard/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: DevOps & Infra
- Author: ZhangHanDong (https://skillmd.com/u/zhanghandong)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/zhanghandong/pipeline-guard

---


# Pipeline Guard

Automatically validates DOT graphs and assigns optimal models before `run_pipeline` executes.

This is a lifecycle hook — it runs transparently before every pipeline execution.

## What it does

1. **Validates DOT structure** — checks for cycles, disconnected nodes, malformed syntax
2. **Assigns models from QoS catalog** — reads `model_catalog.json` scores and assigns:
   - Search workers (dynamic_parallel) → FAST models (round-robin across pool)
   - Analyze/synthesize nodes → STRONG models (best QoS score)
   - Planner model → STRONG
3. **Concurrent distribution** — PID-based random start index so parallel pipelines use different models

## No user action needed

The LLM writes DOT graphs without `model=` attributes. This hook injects them automatically based on live QoS scores.