What these skills are
The whole L&D job, from the training request to the report to executives, as installable Claude skills.
This pack is for L&D managers and specialists, instructional designers, trainers and facilitators, and L&D leaders.
The skills follow learning and development work in the order you do it, in eight stages, from questioning the training request to the one-page report to executives and the plan for the L&D year.
The red line: Claude drafts and checks, people decide: you talk to the learners, the experts and the sponsor, a named person signs off what gets built and what counts as proof, and nothing a learner does in training is used to grade, rank or build a case against them.
Each skill is one action with its named method inside, a fixed output template and a decision a named person makes, and each one points to the next.
No made-up results, retention figures, benchmarks, quotes or returns. Numbers are yours, or a placeholder in brackets. Checks, role-plays, forms and pilots measure the programme, never a learner, a trainer or a manager.
Download all 40 skills. One zip: ready-to-install skill zips, readable SKILL.md files, and the sources behind every method. Free, no signup. Download the zip · Browse on GitHub
The 40 skills you get
The set follows L&D work in eight stages: diagnose the need, work with the experts, design the learning, build and test, run the room, carry it into work, measure what changed, and lead L&D.
1 · Diagnose the need
1. Question the Request
Use when: "We need a training by Friday" lands with the answer already chosen
Output: Intake questions, the business problem in one line, an "is it training?" triage, a reply with a next step and a date
2. Diagnose the Performance Gap
Use when: Every performance problem becomes a training request, whatever the cause
Output: Desired versus actual by task, a cause table, an option per cause, a recommendation that can say "no training"
3. Prepare the Sponsor Conversation
Use when: A course will not fix it and you have to tell the person who asked for the course
Output: One page for the meeting: findings with evidence, the interest behind the request, options, likely objections, shared criteria, the agreement record
4. Assess the Training Needs
Use when: You need to know what a team lacks, not collect a wish list of course titles
Output: Needs by role and task with evidence, a priority table on your criteria, a "training cannot fix" list
5. Map the Actions
Use when: The sponsor wants a course and the expert sends a 147-slide deck
Output: One measurable goal, observable actions, why they are not happening, practice per action, minimum information, a cut list for the expert's material
2 · Work with the experts
6. Interview the Expert
Use when: The expert gives you the standard workflow and solves real cases by intuition
Output: Incident questions in three passes, decision points with the cue the expert noticed, common mistakes, exceptions, scenario seeds
7. Break Down the Task
Use when: The outline follows the manual's chapters and nobody listed what people in the role do
Output: Task list for one role, steps, decisions and standard per task, a difficulty, importance and frequency table, a call per task
8. Plan the Reviews
Use when: Six departments review one course and the revisions never end
Output: Who reviews what at each gate, one named decider, rounds and dates, a feedback log, a change rule for late requests
3 · Design the learning
9. Write the Objectives
Use when: The only objective on the brief is "understand the topic"
Output: Objectives with an observable verb at the level the job needs, a condition and a criterion, rewrites of "awareness only" goals
10. Plan the Evaluation
Use when: Leadership sees a completion rate and nobody agreed what should change
Output: The result in the sponsor's words, leading indicators, behaviours with drivers and owners, learning and reaction checks, a source, owner and date per measure
11. Write the Design Document
Use when: Leaders think AI can do the job and nobody sees the design decisions
Output: Audience by role, strategy, practice and check per objective, every design decision with its rationale, what was left out, a sign-off block
12. Map the Curriculum
Use when: A programme has grown module by module and nothing connects
Output: Modules by objective, practice, check and format, orphan modules and objectives, a simple-to-hard sequence, the learning path
4 · Build and test
13. Write the Scenarios
Use when: New hires finished every module and still cannot make the call on the job
Output: Decision scenarios built from real mistakes, options that tempt, a consequence and feedback per choice, a path map, facts listed for the expert
14. Storyboard the E-Learning
Use when: The build starts from slides and ends as a knowledge dump with voiceover
Output: Screen-by-screen text, visual, narration and interaction, the objective per screen, accessibility notes, a cut list, a build brief for your authoring tool
15. Write the Knowledge Check
Use when: An AI can answer your quiz from a screenshot and it proves nothing
Output: Scenario questions per objective, distractors from real mistakes, feedback for every option, low-stakes retake rules, reporting by item only
16. Build the Role-Play
Use when: People sit through role-play, go out to real customers, and you hope it stuck
Output: Learning points, role cards, a good and a poor model, observer notes that go only to the learner, debrief questions, each learner's own goal
17. Check the AI Draft
Use when: AI drafts the course, AI reviews it, and no person checks that it teaches anything
Output: Every claim traced to an approved source or flagged, alignment row by row, five accessibility checks, a myth and bias scan, the expert sign-off record
18. Run the Learner Walkthrough
Use when: The course looked right in review and the first real learner got lost on screen three
Output: A plan for a handful of volunteers, tasks, think-aloud prompts, an observation sheet about the draft, a ranked issue list, the fix list and the release call
5 · Run the room
19. Plan the Session
Use when: A virtual session competes with email and chat from minute one
Output: A timed flow (open, recall, show, practise, apply), activities per block, a first five minutes on a real problem, slack shown, a transfer close
20. Plan the Group Work
Use when: The plan says "group discussion, 20 minutes" and the opener makes people cringe
Output: One activity card per practice block, an opener built on the work, a virtual variant, a fallback for a quiet room
21. Build the Training Slides
Use when: The deck is the SOP pasted onto slides and the trainer reads it aloud
Output: A slide-by-slide outline tied to the session blocks, one idea per slide, practice slides, what moves to a handout, speaker notes as prompts, a cut list
22. Write the Facilitator Guide
Use when: Another trainer has to run your session without reading slides aloud
Output: Purpose, cue and key points per segment, questions with what a good answer sounds like, activity set-ups, timing flex, four what-if notes
23. Plan the Pilot
Use when: The first full run is also the launch
Output: The questions the pilot must answer, a pass line written in advance, the pilot group, what is observed, the debrief, a change list, a launch, fix or stop call
24. Train the Trainers
Use when: The programme depends on facilitators who each need to learn it first
Output: Delivery learning points, a model run, a teach-back plan, practice rounds that rise in difficulty, delivery notes, a readiness conversation with no score
6 · Carry it into work
25. Build the Job Aid
Use when: People need the answer mid-task, not a 30-minute module
Output: A checklist, decision table or one-page flow chosen by task shape, where it lives in the workflow, the owner, the review date
26. Build the 70-20-10 Plan
Use when: The programme is all classroom and nothing happens back at work
Output: Real assignments per capability, coaching and peer learning tied to them, only the formal pieces they need, owners and check-in dates
27. Space the Practice
Use when: People can recite it in the room and freeze on a live case a month later
Output: Retrieval prompts built on real mistakes, 5-minute micro-units, widening gaps you set, the schedule and channel, results read by item only
28. Brief the Managers
Use when: Learners go back to a manager who never asks what they learned
Output: A before and an after conversation guide, the first on-the-job task, check-in dates in existing meetings, a barrier log about conditions
29. Plan the Onboarding
Use when: Ramp-up takes months and stakeholders blame the talent
Output: A week-by-week path built on the real tasks, practice before each live task, buddy sessions, sign-off by doing the task with a person
7 · Measure what changed
30. Write the Feedback Form
Use when: The smile sheet says everyone loved it and tells you nothing
Output: Four to six items with concrete answer options, open prompts on barriers, an aggregate summary template that rates the programme
31. Check Use on the Job
Use when: The sponsor asks "did anyone actually do anything with it?"
Output: A short anonymous follow-up, the group-level work measures to read beside it, a guide for volunteer conversations, a barrier list, the limits of the evidence
32. Estimate the Training ROI
Use when: The budget holder asks what the programme returned
Output: The chain from application to impact, how the effect was isolated, money conversion with sources, full costs, a low and a high case
33. Report What Changed
Use when: Leaders ask whether it improved anything and the report shows completions
Output: One page for executives, bottom line first: what changed and how sure we are, the chain of evidence, what we cannot claim, the decision asked
34. Review the Programme
Use when: Evaluation data is filed and next year's priorities are set from the inbox
Output: Intended versus actual, why, what to keep and change, a keep, fix or retire call, the items for next year's plan
8 · Lead L&D
35. Plan the L&D Year
Use when: One person must build two programmes at once and every sponsor thinks theirs is first
Output: Programmes ranked against business goals, a capacity and budget line from your own days, a quarterly calendar, a visible not-this-year list
36. Make the Business Case
Use when: You must ask for budget before anything is built and the numbers are guesses
Output: One page in five parts, an assumptions table with low and high values and owners, what would change the answer
37. Brief the Vendor
Use when: The tool was chosen from a feature list and a polished demo
Output: A requirements brief, must-haves against nice-to-haves, a demo script on your own content, weighted selection criteria, the decision record
38. Plan the Build
Use when: The brief lands Thursday and the programme is due Monday
Output: Phases with a deliverable, an owner and a closing decision, review gates, durations from your own projects, a scope line, the risks to the date
39. Plan the Compliance Training
Use when: Compliance training is a box to tick and the quiz record is kept against people
Output: The few decisions each rule turns on, a scenario check per decision, a spaced refresh, a record-use statement, points for a qualified adviser
40. Set the Team AI Rules
Use when: Half the team uses AI quietly, the other half refuses, and nobody tells learners
Output: A use map by task, a never-put-in list, the check a person runs before anything ships, whose name is on the work, a disclosure line per output
Where to start
| Your situation | Skill to run |
|---|---|
| “We need a training by Friday” just landed? | Question the Request |
| Not sure training is the fix? | Diagnose the Performance Gap |
| Have to tell the sponsor it is not a course? | Prepare the Sponsor Conversation |
| The expert gives you the standard workflow, not the real cases? | Interview the Expert |
| Six departments review one course and the revisions never end? | Plan the Reviews |
| The only objective on the brief is “understand the topic”? | Write the Objectives |
| Nobody agreed what should change before the build? | Plan the Evaluation |
| New hires finished every module and still cannot make the call on the job? | Write the Scenarios |
| An AI-made draft that no person has checked? | Check the AI Draft |
| A virtual session competing with email and chat from minute one? | Plan the Session |
| The first full run is also the launch? | Plan the Pilot |
| People need the answer mid-task, not a 30-minute module? | Build the Job Aid |
| Learners go back to a manager who never asks what they learned? | Brief the Managers |
| The sponsor asks if anyone actually did anything with it? | Check Use on the Job |
| Leaders ask whether it improved anything and the report shows completions? | Report What Changed |
| Asking for budget before anything is built? | Make the Business Case |
The quality bar
Every skill in the pack holds the same standard:
- One method, applied properly: each skill uses the real mechanics of its method (the levels, the passes, the gates, the table columns), not a generic “gather, analyse, recommend”
- It tells you where it breaks: every skill says when not to use it and which skill in the guide fits instead
- Same shape every time: a fixed output template, a short “done when” check, and a decision a named person makes
- A person checks before it ships: facts are traced to sources an expert approved, a few learners try the draft, and a pilot runs against a pass line set in advance
- No invented facts: no made-up results, retention figures, benchmarks, quotes or returns. Numbers are yours, or a placeholder in brackets
- Programmes are evaluated, people are not: checks, role-plays, forms and pilots measure the programme, in aggregate, with small groups suppressed. Nothing scores, ranks or tracks a learner, a trainer or a manager
- The red line: Claude drafts and checks, people decide: you talk to the learners, the experts and the sponsor, a named person signs off what gets built and what counts as proof, and nothing a learner does in training is used to grade, rank or build a case against them
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.
This pack is for whoever is accountable for what gets built and what counts as proof. If you run an L&D team or a training practice and want it working this way, with AI carrying the design documents, the storyboards, the checks and the reports so people talk to the learners, the experts and the sponsor and make each call, that is what we build with clients.
I help L&D leaders, instructional designers and trainers build AI-first ways of working, so Claude drafts and checks and people make the calls. Find 30 min together
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 complete-guide-ld-pack@polar-bear-skills
Using Claude on the web instead? Download the zip and upload each skill from its install folder under Customize → Skills.
How to use it: pick the workflow that matches your week
Start with the request: the guide follows the work in order from there. After that, bring the thing that is in front of you.
Answer the request
A training request lands with the answer already chosen.
Question the Request → Diagnose the Performance Gap → Prepare the Sponsor Conversation
Get the real content
The expert sent a 147-slide deck and the standard workflow.
Map the Actions → Interview the Expert → Break Down the Task → Plan the Reviews
Design before you build
The brief says “understand the topic” and nobody agreed what should change.
Write the Objectives → Plan the Evaluation → Write the Design Document → Map the Curriculum
Build and test the draft
Practice and checks to write, then a person and a few learners to try them.
Write the Scenarios → Storyboard the E-Learning → Write the Knowledge Check → Build the Role-Play → Check the AI Draft → Run the Learner Walkthrough
Run the room
A live session to plan, hand to another trainer and pilot before the launch.
Plan the Session → Plan the Group Work → Build the Training Slides → Write the Facilitator Guide → Plan the Pilot → Train the Trainers
Carry it into work
The programme is all classroom and nothing happens back at work.
Build the Job Aid → Build the 70-20-10 Plan → Space the Practice → Brief the Managers
Report and plan the year
Leaders ask what changed and next year’s priorities are due.
Check Use on the Job → Estimate the Training ROI → Report What Changed → Review the Programme → Plan the L&D Year → Make the Business Case
Install it in five minutes
In Claude Code
/plugin marketplace add polar-bear-org/claude-skills /plugin install complete-guide-ld-pack@polar-bear-skills
In Claude.ai or Claude Desktop
- Download the pack and unzip it: its skills folder holds all 40 skills (https://meet-polar-bear.com/skills/complete-guide-ld-pack.zip). In Claude, click your name at the bottom left, then Settings.
- Under Customize, choose Skills. Click Add, then Upload a skill.
- Drag the skill's file into the box, or click to browse. You can add several at once.
- In the pack folder, open skills, then the skill you want, and select its SKILL.md (ldg-question-the-request is a good first one).
- Claude shows the skill's name and what it does. Click Upload.
- The skill opens on its own page. Its toggle should be on.
- Try it in a new chat: type / and the skill's name, then say what you need.
Then try one
Run ldg-question-the-request on the training request you just received: [paste the email here]Run ldg-check-the-ai-draft on this AI-drafted module and the approved procedure: [paste them here]Run ldg-report-what-changed on these figures from the programme: [paste them here]
Questions people ask
What is in the Complete Claude Guide for L&D pack?
40 Claude skills that follow the L&D job in eight stages: diagnose the need, work with the experts, design the learning, build and test, run the room, carry it into work, measure what changed, and lead L&D. Each skill is one action with its named method inside.
How do I install these Claude skills?
In Claude Code, two lines install all 40 at once. In Claude.ai or Claude Desktop, download the zip, open Settings, Skills, Add, and upload the skills you want. Either way it takes about five minutes.
Do I need Claude Code to use them?
No. The skills work in Claude.ai, Claude Desktop and Claude Code. In the app you upload them yourself; in Claude Code one command installs the whole pack.
Do the skills work with ChatGPT or other assistants?
They are built as Claude Skills, and installing them is a Claude feature. Each one is a plain Markdown file, so you can read it and reuse the method elsewhere, but it will not load and trigger by itself outside Claude.
Does Claude grade or rank learners?
No. Claude drafts and checks, people decide. A named person signs off what gets built and what counts as proof, and nothing a learner does in training is used to grade, rank or build a case against them.
Is the pack free?
Yes. It is free to use inside your company and in client work, and not for resale. There is no signup: the download and the install command are on this page.