Agent skill collection for engineers who use Claude Code or Codex, small, composable prompts that enforce TDD, domain modeling, and architecture review.
What it does
Skills for Real Engineers is a curated collection of agent skills by Matt Pocock, built from his own .claude directory, targeting Claude Code, Codex, and other coding agents. The skills are small, composable, and model-agnostic, designed to fix real failure modes in AI-assisted engineering rather th
AI coding agents fail in predictable ways: they misunderstand the task, get verbose, produce untestable code, and quietly rot your architecture. Skills for Real Engineers is a direct antidote, packaging decades of software engineering discipline into slash-commands your agent can actually follow.
Matt Pocock built these skills from his own .claude directory to fix four named failure modes he observed with Claude Code and Codex. Per the README, approaches like GSD, BMAD, and Spec-Kit try to solve the same problem by owning the process entirely, but that trades your control for their bugs. These skills stay small and composable so you can hack them to fit your project.
Skills split on a single axis: user-invoked skills (you type /grill-me, /to-prd) orchestrate a session; model-invoked skills (/tdd, /diagnosing-bugs, /codebase-design) are reusable discipline loops the agent can reach for automatically. A user-invoked skill may call model-invoked ones, but never another user-invoked skill, keeping composition predictable.
/grill-me, /grill-with-docs): a grilling session forces the agent to ask you detailed questions before writing a line./grill-with-docs, /domain-modeling): builds a shared CONTEXT.md glossary so agents use one precise term instead of 20 words./tdd, /diagnosing-bugs): enforces red-green-refactor and a reproduce-minimise-hypothesise-instrument-fix loop./improve-codebase-architecture, /to-prd): scans for deepening opportunities, surfaces them as an HTML report, then grills you through the fix.The /grill-with-docs skill is described in the README as "the single coolest technique in this repo." It runs a grilling session that simultaneously builds a domain model, writes ADRs, and sharpens CONTEXT.md in-place. The payoff compounds: consistent naming, cheaper token spend per session, and a codebase the agent can navigate without re-explaining jargon each time.
These skills require you to stay in the loop. There is no autonomous "run and forget" mode; the grilling sessions and architecture reviews are designed as interactive checkpoints. If you want a fully hands-off agentic pipeline, this is the wrong tool.
Skills for Real Engineers is the clearest implementation of "fundamentals first" AI coding available as a skill pack. One npx command installs the whole set; one /setup-matt-pocock-skills call configures it for your issue tracker. If you are already reaching for Claude Code or Codex daily, this is the obvious add-on to stop vibe-coding drift before it compounds.
Features
Field notes
Reviewed Jun 26, 2026
Best for
Builder outcomes
Watch out
Tested with
Go deeper
Direct skill page for the documentation-grounded grilling workflow.
Skills for Real Engineers is a curated collection of agent skills created by Matt Pocock and published from his personal .claude directory. Each skill is a slash-command that encodes a specific software engineering discipline, from test-driven development to domain modeling to codebase architecture review. The skills are model-agnostic, composable, and MIT licensed, designed to fix the four most common failure modes he observed when using Claude Code, Codex, and similar coding agents.
Run `npx skills@latest add mattpocock/skills` in your terminal, then select the skills and agents you want to install them on, making sure to include `/setup-matt-pocock-skills`. After installation, run `/setup-matt-pocock-skills` inside your agent once per repo; it will ask which issue tracker you use (GitHub, Linear, or local files), what triage labels you apply, and where to save generated docs. The README describes this as a 30-second setup.
Yes, the repository is MIT licensed and free to use. It is installed via a one-line npx command with no paid tier or subscription required. The README mentions a newsletter with around 60,000 subscribers for keeping up with new skills, but that is optional.
It is best for engineers who already use Claude Code or Codex daily and want to stop accumulating quiet architectural debt and misalignment errors. The `/grill-with-docs` and `/grill-me` skills are flagged in the README as the most popular, helping teams build a shared domain language that makes every subsequent agent session cheaper and more precise. The `/improve-codebase-architecture` skill is recommended for running every few days on any codebase built primarily with agents.
The README directly names GSD, BMAD, and Spec-Kit as alternative approaches that try to fix the same problem by owning the entire development process. Pocock's explicit critique is that owning the process removes the developer's control and makes process bugs hard to resolve. Skills for Real Engineers keeps each skill small and independently composable so you stay in control of the flow, while those alternatives hand the workflow to their own orchestration layer.
These skills require an agent that supports slash commands; they cannot be used with a plain chat interface. They are also explicitly interactive: grilling sessions, architecture reviews, and TDD loops are designed as checkpoints that require your participation, so there is no fully autonomous mode. Additionally, the engineering skills require running `/setup-matt-pocock-skills` once per repository before they function, meaning there is a small initial configuration cost for each new project.
Open-source agentic skills framework that brings spec-first, TDD-enforced, subagent-driven software development to Claude Code, Cursor, Codex, and 8 other AI coding harnesses.
Best for Teams wanting a full SDLC enforced automatically on coding agents, from spec through code review39 mental-model skills for Claude Code, first principles, Bayesian, OODA, pre-mortem, and more, all invoke-by-name with no setup.
Best for Engineers, founders, and analysts who want Claude Code to reason through decisions using named, established mental modelsOpen-source collection of reusable agent skills, shared AGENTS.MD rules, and portable helper scripts for Codex and Claude agents, by steipete.
Best for Developers running multiple Codex or Claude agent projects who need one canonical source for shared rules and skillsAgent skill that audits your codebase with a capable model and writes executable plans for cheaper models to implement, by shadcn.
Best for Teams running AI-assisted development at scale who want high-quality code review without paying for a capable model on every execution stepA catalog of repeatable AI agent workflows with built-in feedback loops, plus an installable skill for finding, adapting, and designing your own loops.
Best for Teams using Codex, Cursor, or Claude Code who want agents to iterate reliably instead of making one-shot guessesAnthropic's official skill library for Claude: modular SKILL.md bundles that add repeatable, domain-specific workflows across docs, dev, design, and enterprise tasks.
Best for Teams who need Claude to follow repeatable, specialized workflows across many sessions without re-prompting