# Mlir Development

> Build MLIR or CIR dialects, conversions, passes, and multi-level compiler pipelines. Use when MLIR/CIR is the implementation substrate, including ML and domain-specific compilers.

- Skill: `neversight/mlir-development-2` (Agent Skill, multi-file: 5 files)
- Install (CLI): `npx get-skill add neversight/mlir-development-2`
- Raw SKILL.md: https://api.skillmd.com/api/skills/neversight/mlir-development-2/raw
- Safety review: pending (external: skill-scanner PASS, skillspector PASS)
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: neversight (https://skillmd.com/u/neversight)
- Updated: 2026-09-22
- Page: https://skillmd.com/skills/neversight/mlir-development-2

---


# MLIR development

Define the source and target invariants for each dialect boundary before implementing rewrites or lowering.

## Topic routing

- [Foundations](references/foundations.md) for MLIR's IR model, dialects, operations, types, and attributes.
- [Passes and conversion](references/passes-and-conversion.md) for transformations, patterns, legality, and lowering.
- [Dialects and compilers](references/dialects-and-compilers.md) for built-in dialects, CIR, and ML compiler ecosystems.
- [Testing, tools, and resources](references/testing-tools-and-resources.md) for `mlir-opt`, translation, verification, and upstream references.

Use current dialect definitions and verifier contracts as authority. Test transformations at the narrowest textual or executable boundary that proves legality and semantics.

