What is a skill in Claude?

Think of it like a recipe. Ask any cook to "make dinner" and you'll get something - but probably not the dish you had in mind. Hand them a proper recipe and you get the right result, the same way, every time: the ingredients, the steps, the order, and what it should look like on the plate.

A skill is that recipe for Claude. Claude is the cook. You bring the ingredients - your asset, your failure history, your data. The skill tells Claude exactly how to work with them. (It's just a text file written in markdown - plain text with simple headings - but you don't need to know any of that to use it.)

For Claude's official definition, see What are skills? in the Claude Help Center.

Just a prompt

"Make me dinner."

Claude guesses. You re-explain what you want every single time. The result drifts from one chat to the next.

A skill

"Here's the recipe - follow it."

Claude follows a proven method, asks the right questions, and gives you the same quality output every time.

The difference is your operating context.

A blank prompt doesn't know your plant, your asset hierarchy, or how your team scores criticality and runs FMECAs. A skill comes loaded with that context - the methods, the standards, the questions a senior reliability engineer would actually ask before signing off a recommendation. That's what separates a generic chatbot answer from work that reads like it came from someone who's walked your line and sat in your planning meetings.

And it compounds. Once your site context lives inside the skill, you stop re-explaining it. No more pasting the same CMMS field definitions, the same bad-actor rules, the same "here's how we do FMEAs on this site" every time you open a chat. You write the context once; Claude reuses it on every pump, conveyor, and turnaround scope. That's where the real time savings come from.

The shortcut. Building a skill with your context baked in - your assets, your standards, your way of working - turns Claude from a blank assistant into one that already knows your plant. What used to take a long, repetitive prompt now takes one line: "Run the FMEA skill on this pump."

For reliability & maintenance engineers

The 18 skills you get.

Grouped the way your reliability program actually runs - define the asset first, communicate the decision last - but every skill stands on its own when that's all you need today.

Pillar 1

Asset Foundations

Get the basics right before you analyse anything: what the asset does, how critical it is, and how it's structured.

1.1

Function & Functional Failure Definition

Use whenYou need to define what an asset must do before analysing how it fails.
OutputClear function statements and a structured functional-failure list.
1.2

Criticality Analysis Guide

Use whenYou need to rank assets by consequence and likelihood, defensibly.
OutputCriticality scoring logic and a ranked shortlist to validate.
1.3

Asset Hierarchy & BOM Structuring

Use whenYour asset register or BOM is messy, flat or inconsistent.
OutputA clean parent-child hierarchy and a structured bill of materials.

Pillar 2

Failure Analysis

Understand how and why things fail - from a single event to the assets that bleed your budget.

2.1Reference skill

FMEA Generation

Use whenYou need failure modes, effects and causes for an asset, fast.
OutputA structured FMEA draft, ready for workshop review.
2.2

Weibull & Failure Pattern Analysis

Use whenYou want to read failure history, fit a Weibull, and spot the dominant pattern.
OutputWeibull parameters, pattern read (infant / random / wear-out), and interval implications.
2.3

Root Cause Analysis (RCA) Copilot

Use whenAn event needs a disciplined, evidence-led investigation.
Output5-Why / Ishikawa structure, hypotheses and an evidence plan.
2.4

Bad Actor Analysis

Use whenA handful of assets eat most of your maintenance spend.
OutputA bad-actor shortlist and the angles to attack first.

Pillar 3

Maintenance Strategy

Choose the right tactic for each failure mode, cut what doesn't work - and close the loop when CMMS history tells you the plan needs adjusting.

3.1

Maintenance Strategy Selection

Use whenYou must pick the right task for each failure mode (RCM logic).
OutputAn RCM decision path and a recommended maintenance tactic.
3.2

PM Optimization (Fake-PM Killer)

Use whenYour PM plan is bloated with low-value, calendar-driven tasks.
OutputA PM review with clear keep / cut / merge calls.
3.3

P-F Interval Sizing

Use whenYou need a defensible inspection or monitoring interval.
OutputP-F reasoning and a proposed interval with assumptions.
3.4

Spare Parts & Stock Strategy

Use whenStock levels are guesswork - too much cash or too much risk.
OutputCriticality-based stocking logic and min/max thinking.
3.5

CMMS Data Analysis – Strategy

Use whenYour PM plan and strategy don't match what CMMS history is actually showing on site.
OutputTargeted adjustments to tasks, intervals and tactics - with the work-order evidence behind each call.

Pillar 4

Reliability Modeling & Economics

Frame the hard questions - redundancy, availability and whole-life cost - before you crunch the numbers.

4.1

Reliability Modeling Helper (RBD)

Use whenYou need to frame a reliability block diagram or redundancy question.
OutputModel structure, assumptions and what to compute in your tools.
4.2

Life Cycle Cost (LCC) Analysis

Use whenYou compare options over the full life of an asset.
OutputAn LCC structure with the cost drivers laid out.

Pillar 5

Planning & Turnarounds

Turn decisions into executable work - from a single job package to a full turnaround scope.

5.1

Maintenance Job Planning

Use whenYou turn a task into an executable PMI / SOP / work package.
OutputStep-by-step plan with safety, tools and spares.
5.2

Shutdown / Turnaround Scope Optimization

Use whenA turnaround scope is too big, too vague or unjustified.
OutputA scope challenge with clear in / out / defer logic.

Pillar 6

Metrics & Communication

Turn reliability into metrics that drive action - and recommendations that land with the people who sign off.

6.1

Reliability KPI Architecture

Use whenYour metrics are noisy, lagging or performative.
OutputA decision-linked KPI set that actually drives action.
6.2

Reliability Decision Memo & Storytelling

Use whenA recommendation must land with management, fast.
OutputA one-page decision memo and a clear narrative.

Setup guide

Four steps. Five minutes. Claude on the web.

Step
1

Download the skills

Go to GitHub → github.com/Rob-Reliability/reliability-skills-for-claude, then click Code → Download ZIP.

Prefer one click? Download the ZIP directly.

github.com
Step
2

Create a Claude Project

Go to claude.ai → open Projects → create one, e.g. Reliability Co-pilot.

Claude Code? Put the files in your skills folder and skip steps 2–3.

claude.ai · Projects
Step
3

Upload the skills

Add files → Upload from device, then pick the skill .md files you downloaded from the GitHub repo.

GitHub connected? You can add the skills straight from the repository via Add files → GitHub.

claude.ai · Project knowledge
Step
4

Prompt and call the skill

Start your message by naming the skill you want, then paste in everything you've got on the asset. The skill structures the work — but it can only work with what you give it, so the more context, the better.

The most useful things to drop in:

  • Asset identity — tag, type and duty (e.g. P-101, centrifugal boiler-feed pump).
  • The failure or symptom — what happened, when, and how often.
  • Work-order & maintenance history — repairs, parts replaced, costs, downtime.
  • Condition-monitoring logs — vibration, oil analysis, thermography trends.
  • Operating context & run hours — duty cycle, load, environment.
  • Standards & constraints — anything that applies on your site.

Don't worry about formatting — paste raw exports, tables or notes and let Claude make sense of them. The more data, history and logs you give, the sharper the output. Quality in, quality out.

A few examples:

Run the FMEA Generation skill on this centrifugal pump (P-101, boiler feed). Operating context: continuous duty, 24/7, ambient 35°C. Here are 3 years of work orders, the last vibration report, and the OEM datasheet — use all of it.
Use the RCA Copilot skill on this repeat bearing failure (motor M-204). Below: failure dates, replaced parts, vibration trend before each trip, lube records, and operator notes. Walk me through the likely root causes and what evidence to collect.
Apply the Bad Actor Analysis skill to this 12-month CMMS export (work orders + costs + downtime hours pasted below). Rank the worst assets and tell me where to dig first.

Pick the right skill for your site.

Skip the directory hunt. Find the line that matches the asset, failure or scope you're staring at today.

Define what an asset must do1.1 Function & Functional Failure
Rank assets by criticality1.2 Criticality Analysis
Clean up a messy asset register1.3 Hierarchy & BOM
Build an FMEA fast2.1 FMEA Generation
Fit a Weibull on failure history2.2 Weibull & Failure Pattern
Investigate a failure2.3 RCA Copilot
Find your worst assets2.4 Bad Actor Analysis
Pick the right maintenance task3.1 Strategy Selection
Cut useless PMs3.2 PM Optimization
Tune your plan from CMMS history3.5 CMMS Data Analysis – Strategy
Set an inspection interval3.3 P-F Interval Sizing
Right-size your spares3.4 Spare Parts & Stock
Frame a reliability model4.1 Modeling Helper (RBD)
Compare options over life4.2 Life Cycle Cost
Turn a task into a work package5.1 Job Planning
Challenge a turnaround scope5.2 Turnaround Scope
Fix noisy KPIs6.1 KPI Architecture
Get a recommendation approved6.2 Decision Memo

Skill standards and quality

Each skill is written to push Claude toward the standard a senior reliability engineer would expect before a recommendation leaves the room and lands in front of operations.

Decision-oriented Evidence-aware Honest about AI limits Field-realistic Executive-readable You keep the call

Output quality always follows input quality. Configure the skills with your site context and bring precise asset data, failure history, and standards, and you'll save serious time. Vague inputs, incomplete records, or missing context? The skills structure the work — but they can't do yours for you.

Take them with you

That's the whole toolkit. It's free on GitHub: github.com/Rob-Reliability/reliability-skills-for-claude — click Code → Download ZIP, or download directly. Add a skill to Claude, bring your asset and failure data, keep the decision. If one of them saves you an afternoon on an FMEA or a PM review, that's a win in my book.

This is the first release of an evolving toolkit. The skills are actively maintained and improved — refined with field feedback, updated as methods and standards evolve, with new skills added over time. Star or watch the repo to stay on the latest version, and if you see something that could be sharper, I'd genuinely like to hear it.

You've got the methods — now meet the team. The skills now have a counterpart: Agents for Claude — 7 Claude Code subagents that run these methods as independent specialists, in parallel, in their own clean context: a CMMS digester, a Weibull analyst, a FMECA drafter, an RCA challenger and more. Skills are the recipes; agents are the crew that cooks.

And if you'd rather have these wired into your own CMMS, standards and templates - running on your data instead of a generic prompt - that's literally my day job. Book a 30-minute call, or just email me at hello@robreliability.com and tell me what's eating your team's week.

Hope this helps 🙂 Feel free to reach out if you need anything.

Done for you

Want this applied to your site, done for you?

The same methods, run on your CMMS data by the people who wrote these skills. Start with the Bad Actor Diagnostic – top 10 value destroyers, root causes, costed action plan, in 3–4 weeks. From $7,000.

Bad Actor Diagnostic