Understand any abstract concept through a carefully crafted fable — not by memorizing textbook definitions.
- Background: Inspired by Amanda Askell
- What This Skill Does
- Features
- Quick Start
- Usage Example
- Design Philosophy
- Topics
- Contributing
- License
Amanda Askell is Anthropic's in-house philosopher and head of the Character Alignment team. With a Ph.D. in philosophy from NYU, she is the principal author of Claude's Constitution — the 30,000-word document that shapes Claude's personality and moral compass.
In April 2026, during a podcast interview, Askell shared a small technique that quickly spread through the AI community:
"Ask Claude to write a fable that explains a concept — but never mention the concept's name."
Her core insight is simple yet profound:
"Stories are the most fundamental learning vehicle for humans. Traditional learning tackles definitions head-on. Fables take the opposite path — you get immersed in an engaging story, and only at the very end do you realize what it was about. By then, you've already intuitively grasped the concept's essence."
Askell herself uses this daily: "When I'm bored, I ask Claude to tell me fables. Now I have all these little stories in my head, each one corresponding to a concept from some discipline. Sometimes I can't remember the concept's formal name anymore — but I still remember the story."
This skill takes her idea and expands it from a simple prompt into a complete, structured, reusable methodology.
Askell's original idea is a brilliant starting point. Concept Fable builds on it, systematizing the approach into an 8-step workflow with anti-pattern detection and quality guardrails:
| Askell's Original Idea | This Skill's Systematic Expansion |
|---|---|
| Never mention the concept name | Step 5: Three-act structure with precise reveal-timing control |
| Let the story carry the meaning | Step 4: Causal-chain alignment check — verify metaphor points map to every link of the causal chain before writing; Step 5: the plot's "cause → effect → dilemma" must replicate the concept's operating mechanism |
| — | Step 3: Automatically matches story type to concept characteristics (rivalry → ancient fable / gradual → daily life / emergence → nature / surface vs. reality → dialogue / tool/solution → before-and-after contrast) |
| — | Step 3: Narrative tone matched to the concept's emotional flavor (cautionary / counterintuitive / trade-off / mechanistic) |
| — | Step 7: 7-point self-check — character believability, metaphor accuracy, story independence, reveal timing, conciseness, concreteness, no unexplained jargon |
| — | Anti-pattern table: 6 common failure modes with specific fixes |
| — | "Grandma Test" for metaphor concreteness: the vehicle must be something perceptible in daily life |
| — | Fallback strategies: 4 alternatives when a concept genuinely doesn't fit the fable format |
- 📚 8-Step Structured Workflow: Understand concept → Confirm scope → Choose story type & tone → Validate metaphor mapping → Write fable → Append explanation → Self-check → Output
- 🎭 Automatic Story-Type Matching: Selects the best fable genre based on the concept's core nature (rivalry, gradual emergence, surface-vs-reality contrast, tool/solution)
- 🎯 Causal-Chain Alignment: Goes beyond thematic similarity — the plot's cause-and-effect must mirror the concept's actual mechanism
- 🔀 Before-and-After Contrast: For concepts that exist to solve a problem (e.g., Node.js, caching), the story shows both "life without it" and "life with it," letting the value emerge through the difference
- 🧹 Anti-Pattern Guardrails: Built-in avoidance of 6 failure modes: puppet characters, over-metaphoring, premature reveals, hollow stories, distorted concepts, and metaphors more abstract than the concept itself
- 👵 The Grandma Test: A hard requirement that every metaphor vehicle must use everyday, tangible experiences (cooking, driving, queuing) — no abstract fantasy settings that themselves need decoding
- ✅ 7 Quality Self-Checks: Systematic review after every story; maximum 2 rewrites; honestly report shortcomings if still not satisfied
- 🔄 Feedback Handling Loop: Inaccurate metaphor? Wrong story type? Too subtle or too obvious? Too long or too short? — each feedback type maps to a specific adjustment strategy
- 📖 Standardized Explanation Template: Unified output format with definition, story-to-concept mapping table, prerequisite knowledge notes, and further thinking prompts
What you need: Claude Code installed and running.
Step 1 — Install the skill
This skill requires only a single SKILL.md file to run (English version).
macOS / Linux:
mkdir -p ~/.claude/skills/concept-fable
curl -o ~/.claude/skills/concept-fable/SKILL.md https://raw.githubusercontent.com/hawkongz/concept-fable/main/SKILL.mdWindows (PowerShell):
New-Item -ItemType Directory -Force -Path "$env:USERPROFILE\.claude\skills\concept-fable"
Invoke-WebRequest -Uri "https://raw.githubusercontent.com/hawkongz/concept-fable/main/SKILL.md" -OutFile "$env:USERPROFILE\.claude\skills\concept-fable\SKILL.md"中文用户：请下载中文版 —— 把上面 URL 中的
main/SKILL.md换成main/zh-CN/SKILL.md。
Step 2 — Done.
Restart Claude Code and the skill activates automatically. Try saying:
- "Explain deadlock with a story"
- "Tell me a fable about CAP theorem"
- "Help me understand dependency injection using an everyday scenario"
To update: re-run the commands from Step 1.
Input: "Explain what deadlock is with a story"
Claude will:
- Confirm understanding, map out the causal chain: Thread A holds Lock1 waiting for Lock2 → Thread B holds Lock2 waiting for Lock1 → circular wait → permanent standstill
- Select story type & tone (two-party standoff → ancient fable, cautionary tone)
- Validate metaphor mapping: does every causal chain link have a story beat? (hold-and-wait ✓, circular wait ✓, neither releases ✓)
- Craft a fable that never mentions "deadlock"
- Reveal the concept only at the story's natural conclusion
- Append a professional explanation with a story-to-concept mapping table
- Run the 7-point self-check, then output
The story might be: a sister clutches the salt jar, her brother grips the sugar jar. Her dish needs just a pinch of sugar to finish; his soup lacks only a spoonful of salt. Each holds what the other needs — yet neither will let go. The food goes cold. The soup goes cold.
Traditional teaching follows the path: definition → explanation → example. The reader is a passive recipient.
The fable method reverses this: story → immersion → epiphany → reveal. The reader understands the core of the concept before they even know what they're "learning." The moment of revelation — "oh, that's what this was about" — creates an emotional anchor that makes the understanding far more durable than passive reception.
- Narrative memory advantage: Human brains retain stories far better than abstract definitions. Stories provide context, emotion, and causal chains.
- Understand first, name later: The reader is already grasping the concept through the story; they just don't know its formal name yet. The "aha" moment at reveal is far stronger than being told upfront.
- Lowered cognitive barrier: Everyday scenarios map to abstract concepts — the reader doesn't need to finish a textbook before understanding.
claude-code skill prompt-engineering fable storytelling education learning concept-explanation
Contributions are welcome! If you have ideas for better story types, discover new anti-patterns, or have particularly successful fable examples, please submit an Issue or PR.
MIT © hawkongz
"Sometimes I can't remember the concept's formal name anymore — but I still remember the story." — Amanda Askell