feat: add Claude Code skills exercise modules

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

Author: mjunaidca

EXERCISE-GUIDE.md

@@ -0,0 +1,80 @@
+# 🧠 Building Agent Skills: Complete Exercise Guide
+
+**By Panaversity — Teaching AI Agents to Work YOUR Way**
+
+---
+
+## The Big Idea
+
+A **skill** is a reusable instruction file that teaches an AI agent how to handle a specific type of task consistently. Building skills is the fundamental capability that separates "using AI" from "building with AI."
+
+This guide contains 24 exercises + 3 capstone projects that teach you to build, test, and compose agent skills — without writing any code.
+
+---
+
+## Module Overview
+
+| Module | Focus | Key Skill Learned |
+|--------|-------|-------------------|
+| 1 | Understanding Skills | Reading and analyzing existing skills |
+| 2 | First Skills | Writing simple, single-purpose SKILL.md files |
+| 3 | Skills with Examples | Using examples to constrain and improve output |
+| 4 | Skills with References | Referencing external documents and standards |
+| 5 | Testing & Iteration | Systematic evaluation and improvement |
+| 6 | Composing Skills | Chaining skills into workflows |
+| 7 | Real-World Skills | Production-ready skill building |
+| 8 | Capstone | Complete skill suites for real scenarios |
+
+---
+
+## The Skill-Building Framework
+
+Use this for EVERY skill you build:
+
+### 1. DEFINE — What problem does this skill solve?
+- What task is being automated?
+- What does "good" output look like?
+- What's the current pain (without the skill)?
+
+### 2. DRAFT — Write the first version of SKILL.md
+- Frontmatter (name + trigger description)
+- Step-by-step process
+- Output format
+- Rules and constraints
+
+### 3. TEST — Run it on real examples
+- Start with 2-3 "normal" test cases
+- Then try edge cases designed to break it
+
+### 4. EVALUATE — Score the output
+- Does it match the expected format?
+- Is the content correct?
+- Would you actually USE this output?
+
+### 5. IMPROVE — Fix what's broken
+- Add examples for areas that were inconsistent
+- Add rules for edge cases that weren't handled
+- Tighten vague instructions
+
+### 6. REPEAT — Until quality is consistent
+- Test → Evaluate → Improve is an iterative loop
+- 2-3 rounds is typical; production skills may need 5+
+
+---
+
+## Self-Assessment Rubric
+
+| Criteria | 1 (Beginner) | 2 (Developing) | 3 (Proficient) | 4 (Advanced) |
+|----------|:---:|:---:|:---:|:---:|
+| **Trigger Description** | Too vague or too broad | Covers main cases | Specific and complete | Handles edge cases in trigger |
+| **Instructions** | Missing steps | Basic steps listed | Clear, ordered, complete | Includes decision logic |
+| **Examples** | None | One embedded example | Multiple examples + anti-examples | Example pairs showing input→output |
+| **Edge Case Handling** | Not considered | Some mentioned | Rules for common edges | Tested and verified |
+| **Testing** | Ran once | Tested 2-3 cases | Systematic test suite | Tested by another person |
+| **Iteration** | First draft only | One revision | Multiple iterations | Measurably improved via rubric |
+
+---
+
+*Built for Panaversity's AI-Native Development Curriculum*
+*These exercises work with Claude Code (terminal) or Cowork (desktop app).*
+*No programming knowledge required — just clear thinking and iterative refinement.*

README.md

@@ -0,0 +1,129 @@
+# 🧠 Building Agent Skills: A Practical Exercise Guide
+
+**By Panaversity — Teaching AI Agents to Work YOUR Way**
+
+---
+
+## What Are Agent Skills?
+
+A **skill** is a set of instructions you write that teaches Claude *how* to handle a specific type of task. Instead of writing a long prompt every time, you write the skill once — and Claude follows it automatically whenever that type of task comes up.
+
+Think of it like this:
+- **Without a skill:** You explain your formatting preferences, writing style, data rules, and output requirements every single time.
+- **With a skill:** You write those instructions once in a `SKILL.md` file, and Claude follows them automatically forever.
+
+Skills are the bridge between "using AI" and "building AI systems." They're how you go from one-off prompting to **reusable, consistent, high-quality AI workflows.**
+
+### Anatomy of a Skill
+
+Every skill lives in a folder and has at minimum one file:
+
+```
+my-skill/
+├── SKILL.md          ← The brain — instructions Claude follows
+├── examples/          ← Optional: sample inputs/outputs Claude learns from
+├── templates/         ← Optional: templates for consistent output
+└── references/        ← Optional: reference material Claude can consult
+```
+
+The `SKILL.md` file has three parts:
+1. **Frontmatter** — Name and description (tells Claude WHEN to use the skill)
+2. **Instructions** — Step-by-step guidance (tells Claude HOW to do the task)
+3. **Examples/Rules** — Concrete examples and constraints (tells Claude WHAT good output looks like)
+
+### Where Skills Live
+
+- **Claude Code:** In your project's `.claude/skills/` folder, or any folder Claude can access
+- **Cowork:** As plugins you install in the Claude Desktop app's Cowork tab
+
+---
+
+## 📂 Package Structure
+
+```
+skills-exercises/
+├── README.md                          ← You are here
+│
+├── module-1-understanding-skills/     ← Read & analyze existing skills
+│   ├── exercise-1.1-anatomy/          (Dissect a skill file)
+│   ├── exercise-1.2-when-to-skill/    (Identify skill-worthy tasks)
+│   └── exercise-1.3-skill-vs-prompt/  (Compare skill vs raw prompt)
+│
+├── module-2-first-skills/             ← Build simple, single-purpose skills
+│   ├── exercise-2.1-email-style/      (Writing style skill)
+│   ├── exercise-2.2-file-organizer/   (File organization rules)
+│   └── exercise-2.3-data-cleaner/     (Data standardization skill)
+│
+├── module-3-skills-with-examples/     ← Skills that learn from examples
+│   ├── exercise-3.1-report-formatter/ (Report with example outputs)
+│   ├── exercise-3.2-meeting-minutes/  (Meeting notes with templates)
+│   └── exercise-3.3-feedback-writer/  (Feedback with tone examples)
+│
+├── module-4-skills-with-references/   ← Skills with reference material
+│   ├── exercise-4.1-brand-voice/      (Brand guidelines skill)
+│   ├── exercise-4.2-policy-checker/   (Policy compliance skill)
+│   └── exercise-4.3-curriculum-skill/ (Teaching standards skill)
+│
+├── module-5-testing-and-iteration/    ← Test, evaluate, improve skills
+│   ├── exercise-5.1-edge-case-hunt/   (Find where skills break)
+│   ├── exercise-5.2-before-after/     (Measure skill improvement)
+│   └── exercise-5.3-user-testing/     (Test with classmates)
+│
+├── module-6-composing-skills/         ← Combine skills for workflows
+│   ├── exercise-6.1-pipeline/         (Chain skills together)
+│   ├── exercise-6.2-skill-library/    (Build a personal skill collection)
+│   └── exercise-6.3-team-skills/      (Skills for team workflows)
+│
+├── module-7-real-world-skills/        ← Build production-ready skills
+│   ├── exercise-7.1-invoice-processor/(Business process skill)
+│   ├── exercise-7.2-content-pipeline/ (Content creation skill)
+│   └── exercise-7.3-research-analyst/ (Research workflow skill)
+│
+└── module-8-capstone/                 ← Choose one capstone project
+    ├── capstone-A-business-ops/       (Complete business operations suite)
+    ├── capstone-B-education-kit/      (AI-native course delivery system)
+    └── capstone-C-personal-ai/        (Personal productivity skill set)
+```
+
+---
+
+## 🚀 How to Use These Exercises
+
+### With Claude Code
+1. Navigate to the exercise folder
+2. Claude Code will automatically read SKILL.md files when they're in your project
+3. Test skills by giving Claude tasks that should trigger them
+
+### With Cowork
+1. Point Cowork at the exercise folder
+2. Ask Claude to read the SKILL.md and follow its instructions
+3. For plugin-style skills, install them via the Cowork Plugins panel
+
+### The Learning Loop
+Every exercise follows this cycle:
+1. **Read** the existing SKILL.md (or starter)
+2. **Test** it with the provided sample tasks
+3. **Identify** what's missing or wrong
+4. **Improve** the skill
+5. **Re-test** and compare results
+6. **Reflect** on what made the skill better
+
+---
+
+## 📋 Recommended Schedule
+
+| Week | Module | Focus |
+|------|--------|-------|
+| 1 | Module 1 | Understanding what skills are |
+| 2 | Module 2 | Writing your first skills |
+| 3 | Module 3 | Adding examples to skills |
+| 4 | Module 4 | Adding reference materials |
+| 5 | Module 5 | Testing and improving skills |
+| 6 | Module 6 | Composing skills into workflows |
+| 7 | Module 7 | Building real-world skills |
+| 8 | Module 8 | Capstone project |
+
+---
+
+*Built for Panaversity's AI-Native Development Curriculum*
+*No coding required — just clear thinking and iterative refinement.*

module-1-understanding-skills/exercise-1.1-anatomy/INSTRUCTIONS.md

@@ -0,0 +1,37 @@
+# Exercise 1.1 — Anatomy of a Skill
+
+## Goal
+Learn to read and understand skill files by dissecting three real skill examples of increasing complexity.
+
+## What You Have
+Three skill folders in `sample-skills/`:
+1. `simple-greeting/` — A tiny skill (just formatting rules)
+2. `weekly-report/` — A medium skill (with templates)
+3. `content-reviewer/` — A complex skill (with examples and references)
+
+## Your Tasks
+
+### Task A: Read and Annotate
+Open each SKILL.md file and identify these parts:
+1. **Frontmatter** — The `---` block at the top (name, description)
+2. **Trigger conditions** — When should Claude use this skill?
+3. **Step-by-step instructions** — What does Claude do?
+4. **Output format** — What does the final output look like?
+5. **Rules/constraints** — What must Claude always or never do?
+
+For each skill, write a 3-sentence summary: What it does, when it triggers, and what makes the output good.
+
+### Task B: Predict the Output
+In `test-prompts/`, you'll find sample user prompts for each skill. Before running them:
+1. Read the prompt
+2. Read the skill
+3. Write down what you THINK Claude will produce
+4. Then run the prompt and compare
+
+### Task C: Find the Gaps
+Each skill has an intentional weakness. Find it:
+- `simple-greeting/` — What happens if the user wants a greeting in a non-English language?
+- `weekly-report/` — What if a section has no data?
+- `content-reviewer/` — What if the content is an image, not text?
+
+Write one sentence describing each gap and how you'd fix it.

module-1-understanding-skills/exercise-1.1-anatomy/sample-skills/content-reviewer/SKILL.md

@@ -0,0 +1,93 @@
+---
+name: content-reviewer
+description: Reviews written content (blog posts, articles, documentation, marketing copy) against a quality checklist and provides structured feedback. Use when the user asks to review, critique, edit, or improve any written content.
+---
+
+# Content Quality Reviewer
+
+## When to Use
+- User asks to "review," "critique," "edit," or "improve" written content
+- User shares a document and asks "what do you think?" or "any feedback?"
+- User wants a quality check before publishing
+
+## Review Process
+
+### Step 1: Identify Content Type
+Determine what kind of content this is:
+- Blog post / article
+- Technical documentation
+- Marketing copy
+- Internal communication
+- Educational material
+
+### Step 2: Read the Full Content
+Read everything before commenting. Never start feedback mid-way through.
+
+### Step 3: Score Against Checklist
+Rate each dimension 1-5 and provide specific evidence:
+
+| Dimension | What to Check |
+|-----------|---------------|
+| **Clarity** | Can a reader understand the main point in the first 2 paragraphs? |
+| **Structure** | Does it flow logically? Are transitions smooth? |
+| **Audience Fit** | Is the language appropriate for the target reader? |
+| **Actionability** | Does the reader know what to do after reading? |
+| **Conciseness** | Could any section be cut without losing meaning? |
+| **Evidence** | Are claims supported with data, examples, or sources? |
+
+### Step 4: Generate Feedback Report
+
+Use this structure:
+
+```markdown
+# Content Review: [Title]
+
+## Overall Assessment
+[2-3 sentences: What's the piece about, who's it for, and is it ready?]
+
+## Scorecard
+| Dimension | Score | Key Finding |
+|-----------|-------|-------------|
+| Clarity | X/5 | [one sentence] |
+| Structure | X/5 | [one sentence] |
+| Audience Fit | X/5 | [one sentence] |
+| Actionability | X/5 | [one sentence] |
+| Conciseness | X/5 | [one sentence] |
+| Evidence | X/5 | [one sentence] |
+
+**Overall: X/30**
+
+## Top 3 Strengths
+1. [Specific strength with quote or reference]
+2. [Specific strength]
+3. [Specific strength]
+
+## Top 3 Improvements
+1. [Specific issue + suggested fix]
+2. [Specific issue + suggested fix]
+3. [Specific issue + suggested fix]
+
+## Line-by-Line Notes
+[Only if user requests detailed feedback]
+```
+
+### Step 5: Offer Next Steps
+After delivering the review, ask:
+- "Want me to implement these suggestions?"
+- "Want me to focus on rewriting any specific section?"
+- "Want a revised version with all changes applied?"
+
+## Rules
+- NEVER just say "this is good" — always find at least 2 specific improvements
+- NEVER rewrite the content unless asked — the review should respect the author's voice
+- Feedback must be specific: "Paragraph 3 is vague" is bad; "Paragraph 3 claims 'significant growth' without a number — add the percentage" is good
+- Score honestly. A 5/5 means genuinely excellent, not just "no obvious problems"
+- Reference specific sentences/paragraphs in feedback using quotes
+
+## Examples of Good vs Bad Feedback
+
+**Bad:** "The introduction could be stronger."
+**Good:** "The introduction buries the main point. The key insight — that AI agents reduce SaaS costs by 30% — doesn't appear until paragraph 4. Lead with it."
+
+**Bad:** "Good use of examples."
+**Good:** "The Stripe integration example in section 3 perfectly illustrates the concept for a technical audience. More examples like this throughout would strengthen the piece."

module-1-understanding-skills/exercise-1.1-anatomy/sample-skills/simple-greeting/SKILL.md

@@ -0,0 +1,41 @@
+---
+name: simple-greeting
+description: Generates professional greeting messages for emails, Slack, and meeting openers. Use when the user asks to write a greeting, welcome message, or opening line for any professional communication.
+---
+
+# Professional Greeting Generator
+
+## When to Use
+- User asks for a greeting, welcome message, or icebreaker
+- User is writing an email or message opening
+- User needs to introduce themselves or their team
+
+## Instructions
+
+1. Ask (or infer from context) the **channel**: email, Slack, meeting, or presentation
+2. Ask (or infer) the **relationship**: first contact, existing colleague, manager, client, or external partner
+3. Ask (or infer) the **tone**: formal, friendly-professional, or casual
+
+### Generate the greeting following these rules:
+
+**Email Greetings:**
+- First contact: "Dear [Name]," + one sentence of context for why you're reaching out
+- Existing colleague: "Hi [Name]," + one warm sentence (reference a recent interaction if possible)
+- Client/external: "Dear [Name]," + one sentence acknowledging the relationship
+
+**Slack Messages:**
+- Always start with "Hey [Name]!" or "Hi [team/channel]!"
+- Keep the opener to one line maximum
+- Never use "Dear" in Slack
+
+**Meeting Openers:**
+- Start with "Good [morning/afternoon], everyone."
+- Add one sentence of context: what the meeting is about
+- Keep it under 30 words
+
+## Rules
+- Never use "I hope this email finds you well" — it's overused
+- Never use emojis in emails to clients or first contacts
+- Slack messages MAY use 1-2 emojis if the tone is casual
+- Always match the formality level of the context
+- If the user doesn't specify a name, use [Name] as a placeholder