30 Claude Skills for L&D and Corporate Training

What these skills are

One piece of L&D work, from the request to the evidence, as installable Claude skills. The skills follow the job in the order it happens. You find the real need before anyone says "course", get the know-how out of the experts, design the programme, build the materials, deliver it live, make it stick back at work, prove what changed, and run the L&D desk with the time and budget you actually have.

The red line: Claude drafts the learning and an expert signs it off. People learn by practising with other people, and no quiz, survey or role-play is ever used to grade, rank or build a case against a person.

Each skill is one named method or one artifact you walk away with: an action map, an SME interview guide, a design document the sponsor signs, a branching scenario, a job aid, a facilitator guide, a Kirkpatrick evaluation plan, a conservative ROI range. Each works from what you paste — the request email, your notes from the SME call, the slide deck you were sent, the figures finance gave you — with no LMS connector to install, and each one ends with the decision a named person makes, by a date. Legal, regulatory and accessibility-compliance points are marked "check with a qualified adviser" rather than answered.

Mechanically, each skill is one folder with a SKILL.md file. The 30 are grouped into eight stages of the work, and each one gives full value on its own: run one, run a stage, or work through the lot in the order the job moves.

Download all 30 skills. One zip: ready-to-install skill zips, readable SKILL.md files, and the evidence notes behind every method. Free, no signup. The same 30 skills are open on GitHub: polar-bear-org/claude-skills.

The 30 skills you get

The set follows the work, in eight stages: finding the real need, getting it out of the experts, designing the programme, building the materials, delivering it live, making it stick, proving what changed, and running the L&D desk.

1 · Find the real need

1. Training Request Intake

Use when: "We need a training by Friday" lands with the answer already chosen

Output: The intake questions, the business problem in one line, an "is it training?" triage, and a reply to the requester

2. Performance Gap Analysis

Use when: The manager, process or tools may be the cause, not a skill gap

Output: Desired versus actual performance, a cause table that separates skill and knowledge from information, tools, incentives and process, the non-training options, and a recommendation

3. Training Needs Assessment

Use when: You need what a team actually lacks, not a yearly wish list

Output: Needs by role and task with the evidence for each, a priority table, and a clear split of what training can and cannot fix

4. Action Map

Use when: The sponsor wants a course and you need "they will do", not "they should know"

Output: One measurable goal, the observable actions, why they are not happening, the practice activities, and the minimum information people need

2 · Get it out of the experts

5. SME Interview Guide

Use when: The expert gives you the standard workflow but solves real cases by intuition

Output: An incident-based interview plan, then the notes structured into decision points, a common-mistakes list, exceptions and scenario seeds

6. Content Triage

Use when: The SME says learners need to know everything and sends the whole slide deck

Output: A must-do, must-know, look-up and cut table, the job aid candidates, and a cut list with reasons the SME can accept

7. SME Review Plan

Use when: Six departments review one course and the revisions never end

Output: Who reviews what at each gate, one named decider, the rounds and dates, a feedback log, and a change rule for late requests

3 · Design the programme

8. Learning Objectives

Use when: The only objective is "understand the topic"

Output: Observable verbs at the right Bloom's level, a condition and a criterion for each objective, and rewrites of "awareness only" goals

9. Instructional Design Document

Use when: Nobody sees the design decisions behind a course

Output: The audience and need, the strategy, practice and a check per objective, the evaluation, every design decision with its rationale, and a sign-off block

10. Curriculum Map

Use when: A programme has grown module by module and nothing connects

Output: Modules by objective, activity, check and format — live, e-learning, on the job — a sequence from simpler to harder whole tasks, and a learning path

4 · Build the materials

11. Scenario-Based Learning

Use when: People finished every module and still cannot make the call on the job

Output: Decision scenarios drawn from real mistakes, branches with their consequences, and feedback for every choice

12. E-Learning Storyboard

Use when: The build starts from slides and ends as a knowledge dump with voiceover

Output: Screen-by-screen text, visuals, interactions and narration, accessibility notes, and the objective each screen serves

13. Knowledge Check

Use when: An AI can answer your quiz from a screenshot and it proves nothing

Output: Scenario questions per objective, distractors built from real mistakes, feedback for every option, and low-stakes retake rules

14. Job Aid

Use when: People need the answer mid-task, not a 30-minute module

Output: A checklist, decision table or one-page flow, where it lives in the workflow, and who updates it and when

15. AI Content QA Checklist

Use when: AI drafted the course, AI reviewed it, and no human checked it teaches anything

Output: Accuracy traced to an approved source, objective alignment, accessibility checks, a myth and bias scan, the update owner, and the SME sign-off record

5 · Deliver it live

16. Session Plan

Use when: A virtual session competes with email and chat from minute one

Output: A timed flow — open, recall, show, practise, apply — activities and materials per segment, and a scripted first five minutes

17. Facilitator Guide

Use when: Trainers are told to read the SOP aloud from slides

Output: Cues and key points per segment, questions to ask with what a good answer sounds like, activity setup instructions, timing flex, and what-if notes

18. Role-Play Training

Use when: People sit through role-play, meet real customers, and you hope it stuck

Output: Learning points, role cards, a good and a poor model example, observer notes that go only to the learner, and debrief questions

19. Train-the-Trainer Kit

Use when: The programme depends on facilitators who each need to learn it first

Output: A teach-back plan, practice rounds that rise in difficulty, delivery notes per module, and a readiness conversation guide with no score

6 · Make it stick

20. 70-20-10 Plan

Use when: The programme is all classroom and nothing happens back at work

Output: On-the-job assignments, coaching and peer learning, and the formal pieces, for a role or a programme

21. Spaced Repetition Plan

Use when: People forget a 60-minute course within weeks

Output: Retrieval prompts at widening gaps, 5-minute micro-units, and the schedule and channel

22. Manager Follow-Up Kit

Use when: Learners go back to a manager who never asks what they learned

Output: Before and after conversation guides, the first on-the-job task, check-in dates, and a barrier log

23. Onboarding Training Plan

Use when: Ramp-up drags on and stakeholders blame the talent

Output: A week-by-week learning path, the real tasks a new hire must handle, buddy sessions, and sign-off by doing the task with a person

7 · Prove what changed

24. Kirkpatrick Evaluation Plan

Use when: Leadership sees a high completion rate and moves on

Output: The targeted Level 4 result and its leading indicators, the critical behaviours and required drivers, the Level 2 and Level 1 checks, and a data source, owner and date for every measure

25. Training Evaluation Form

Use when: The smile sheet says everyone loved it and tells you nothing

Output: Questions on relevance, confidence to apply and after-training support, open prompts on barriers, and an aggregate summary that rates the programme, never the trainer

26. Success Case Study

Use when: You need evidence of use on the job, not more survey scores

Output: A short survey on where the training was used and where it stalled, an interview guide for volunteers, anonymous case write-ups with evidence, and a barrier list across cases

27. Training ROI Estimate

Use when: The budget holder asks what the programme returned

Output: The benefits and how the programme's effect was isolated, the money conversion with a source for every value, the full costs, and a conservative ROI range with caveats and intangibles

8 · Run the L&D desk

28. Training Plan

Use when: One person must build two programmes at once

Output: Programmes ranked against business goals, a capacity and budget check, a quarterly calendar, and a not-this-year list

29. ADDIE Project Plan

Use when: The brief lands Thursday and the programme is due Monday

Output: Phases with deliverables and owners, review gates, durations from your own past projects, SAM iterations where they fit, and a scope line for the deadline

30. Compliance Training

Use when: Compliance is a box to tick and quiz records are kept as evidence against people

Output: The few decisions that matter per rule, scenario checks with feedback, a refresh cadence, and a record-use statement, with legal points marked "check with a qualified adviser"

Where to start, and what the pack rests on

There is no compulsory sequence. You bring one live piece of work — the email asking for a negotiation course by Friday, your notes from twenty minutes with the SME, the slide deck you were told to turn into e-learning — and open the skill that matches where it is stuck. The order in the catalogue is the order the work moves in: the real need first, because a request nobody diagnosed is what makes every later step wasted effort, then getting the know-how out of the experts, which is what decides whether the practice looks like the job.

Most of them chain. ld-training-request-intake turns a request into a stated business problem, ld-performance-gap-analysis says whether training is the fix at all, and ld-action-map turns what is left into the actions people must take. ld-sme-interview-guide and ld-content-triage produce the raw material, ld-learning-objectives and ld-design-document fix what people will be able to do and get it signed off, and ld-curriculum-map sequences it. Building runs from there: ld-scenario-based-learning, ld-elearning-storyboard, ld-knowledge-check and ld-job-aid make the materials, and ld-ai-content-qa is the gate before a learner sees any of it. ld-session-plan and ld-facilitator-guide put it in a room, ld-70-20-10-plan, ld-spaced-repetition-plan and ld-manager-follow-up-kit carry it back to the work, and ld-kirkpatrick-evaluation-plan, planned before launch, tells ld-training-evaluation-form, ld-success-case-study and ld-training-roi what to look for.

The pack is research-informed, not validated. A scan of what L&D people say they struggle with set the order of the skills and the words in each "use when": 1,019 public items collected in late September 2026, of which 206 pain statements were kept and grouped into thirteen struggles. Reddit leans to instructional designers and venting, LinkedIn leans to L&D leaders and consultants, corporate trainers are under-represented, and relevance search is not a random sample, so the counts show emphasis in what was pulled, not how common a problem is across the profession. The methods themselves come from their originators' pages, institution guides and peer-reviewed papers — Kirkpatrick Partners on the four levels and planning from Level 4 down, the ROI Institute on isolating a programme's effects and converting them to money, Brinkerhoff (2005) on the Success Case Method, Work-Learning Research's LTEM on what attendance and satisfaction cannot tell you, the Center for Creative Leadership on 70-20-10 as a guideline rather than a formula, CIPD on learning needs analysis, Roediger and Karpicke (2006) with Dunlosky and colleagues (2013) on retrieval practice, Cepeda and colleagues (2006) on spacing, Blume and colleagues (2010) on transfer and the work environment, Taylor and colleagues (2005) on behaviour modelling, Merrill (2002) on first principles of instruction, Sweller and colleagues (2019) on cognitive load, Bauer (2010) for the SHRM Foundation on onboarding, W3C's WCAG 2.2 on accessibility, Cathy Moore on action mapping, Hoffman and colleagues (1998) on the critical decision method, plus revised Bloom's taxonomy, backward design, Gagné's nine events, and ADDIE with Allen Interactions' SAM. The one shared file in the pack, resources/evidence-and-sources.md, lists them source by source with the finding used and the limit on it. Most of that learning research was done with students or in short lab tasks and the transfer research is largely correlational, so the pack treats it as direction for design. No skill here has been tested in a trial.

Setup guide

1. Download the pack. One zip: an install folder with 30 ready-to-upload skill zips, a skills folder with the same 30 skills as readable SKILL.md files, and the evidence notes — resources/evidence-and-sources.md — which name the source behind every method and where it stops.
2. Install your skills. In Claude Code, add the marketplace and install ai-for-training-pack, and all 30 load at once. In Claude, turn on code execution in Settings, Capabilities, then go to Customize, Skills and upload one zip per skill from the install folder. Prefer working from files? Add the SKILL.md files to your Project knowledge instead; it works, just less cleanly.
3. Bring one live request. No connector, no LMS integration, no admin: every skill works from what you paste — the email asking for a course by Friday, your notes from twenty minutes with the SME, the slide deck you were told to turn into e-learning. Start with the request on your desk and write "run ld-training-request-intake". Where you have little, each skill starts from the minimum and marks the result as a first draft.

Where to start

The quality bar

Every skill in the pack holds the same standard, the one we hold when we train our own people:

• Every skill asks what you have first, and works from a minimum when you have little, marking the result as a first draft
• Every method is named and attributed to its originator through a public source, and its real mechanics drive the steps
• Every output has a fixed shape and ends with the decision a named person makes, and by when
• No invented numbers: thresholds, costs and targets stay as [placeholders] until you set them
• Needs are analysed at role and task level, never by rating named people, and survey results are reported in aggregate with small groups suppressed
• Legal, regulatory and accessibility-compliance points are marked "check with a qualified adviser", never stated as advice
• The red line: Claude drafts the learning and an expert signs it off — people learn by practising with other people, and no quiz, survey or role-play is ever used to grade, rank or build a case against a person

Who made this

Polar Bear is a people ops consultancy for human-size teams (20 to 200 people). Built by ex-McKinsey founders with a dream to make AI work for People, not instead of them. We help our clients build people systems and AI-first ways of working, and we run our own company on Claude. This pack is the free, self-serve version of how we work.

The pack carries one L&D desk's work. When you want your whole team working this way, AI carrying the overhead so people do the part only people can do, across hiring, management and everyday operations, that's what we build with clients.

Meet Pauline. A training request with no diagnosis, or a programme nobody applies on Monday? Bring the request and the sponsor who already picked the answer. Book a 30-minute call · Pauline on LinkedIn

Install in Claude Code

Two lines, and every skill in the pack loads at once.

/plugin marketplace add polar-bear-org/claude-skills
/plugin install ai-for-training-pack@polar-bear-skills

Using Claude on the web instead? Download the zip and upload each skill from its install folder under Customize → Skills.