feynman-explain
Fixed-format explanation discipline for concepts, systems, and code.
Skill by ThodsaphonSonthiphin
Install
Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.
npx agentmarket install skill/thodsaphonsonthiphin-feynman-explain
Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.
Get the weekly diff
The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.
About
Reviews
Write a reviewNo reviews yet. Installed it? Reviews come from verified installers only. How reviews work
Permissions
What this skill can do and what it asks for, from its SKILL.md. Checked Oct 25, 2026.
- Tools
- No tools found
- Secrets
- None asked for
- Install scripts
- None found
Network hosts it talks to 1
codeload.github.com· found in install command
Files it touches
~/.claude/skills/feynman-explain
What's inside
- Skill name
feynman-explain- Description
- Fixed-format explanation discipline for concepts, systems, and code. Use when the user asks "อธิบาย…", "นี่คืออะไร", "อันนี้ทำงานยังไง", "explain X", "what is X", "how does X work", "why does this exist"; when they ask for a simpler, shorter, or plainer version of something; when they say an answer was too long, unformatted, or unreadable; or when they invoke /feynman. Also use before teaching, onboarding, or presenting a concept to someone else.
- License
- MIT
- Instructions
- 144 lines in SKILL.md · view on GitHub
Files (2)
- SKILL.md
- diagram-convention.md · references/
SKILL.md (first lines)
# Feynman Explain An explanation you cannot compress is an explanation you do not understand yet. The answer is a **fixed card with hard caps**: the fuzzy parts get named out loud, the gaps get filled from real evidence, and what survives is short enough to read in one screen. ## The output contract Emit these parts, in this order, and nothing else. Write the headings in the user's language. | Slot | ไทย | English | Cap | |---|---|---|---| | ① | `① อธิบายง่าย ๆ` | `① Plain explanation` | ≤ 60 words + one analogy line | | ② | `② จุดที่ยังคลุมเครือ` | `② Fuzzy parts` | table, ≤ 3 gap rows + one jargon row per term used | | ③ | `③ เติมช่องว่าง` | `③ Gaps filled` | ≤ 25 words per gap (question → answer → simple version → source) | | ④ | `④ อธิบายใหม่` | `④ Tightened explanation` | ≤ 80 words + exactly 3 takeaways | | 🎯 | `🎯 30 วิ` | `🎯 30 seconds` | one sentence, quoted | | → | `→ ต่อ:` | `→ Next:` | one line offering the next move | Shape of ② and ③: ``` ② จุดที่ยังคลุมเครือ | จุดที่พูดคลุม | คำถามจริงที่ค้าง | | "..." | ... | | jargon | อธิบายง่ายได้? | | "..." | ได้ / ไม่ได้ / ครึ่ง ๆ | ③ เติมช่องว่าง Gap 1 — <คำถาม> → <คำตอบ> → พูดง่าย ๆ ว่า <…> (ที่มา: path/file.cs:118) ``` **Total default budget: ≤ 250 words.** Over budget means **cut scope** — narrow the topic, drop the weakest gap, shorten the analogy. Never add a section, never spill into extra prose, never continue past `→ ต่อ:`. No preamble, no "great question", no closing summary paragraph — 🎯 *is* the summary. Do not emit a process diagram of this skill: it costs screen space the answer needs. ## Language Mirror the language the user asked in. Keep the technical term in English on first use with a ≤ 8-word gloss in parentheses — `pipeline (ลำดับโค้ดที่วิ่งต่อกัน)` — then use the English term from there on. ## ② is not theater Name at least one real fuzzy point in your own ① — the place you hand-waved, the word you leaned on. Every piece of jargon that appeared in ① gets a jargon row with an honest verdict. If you genuinely have no fuzzy point, say so and give the reason instead of inventing one. ## ③ fills gaps from evidence, not confidence | Gap is about | Fill it from | Cite | |---|---|---| | the user's system — code, repo, D365/Dataverse org, DB, pipeline, config | open the file / run the query / check git **before** answering | `path/file:line`, the query, or the commit | | an external product, API, or platform behaviour | official docs (the `microsoft-docs` skill, or a fetched vendor page) | the URL | | vendor lifecycle — SKU, exam code, retirement, pricing | hand off to `verify-then-advise` | that skill's verdict | | a general concept with no artifact behind it | your own knowledge | mark it `(จากความรู้)` / `(from knowledge)` | A gap you cannot fill **stays in ②** as `ยังไม่ชัด`. An unfilled gap is a finding you report, not a failure you paper over — and never a sentence written confidently enough to hide it. ## Analogy rule One line, concrete, from everyday life: `X เหมือน Y เพราะ Z`. "Git is like time travel" is abstract and fails; "Git เหมือนการเซฟดราฟต์เรียงความ — ย้อนกลับไปดราฟต์ไหนก็ได้" works. ## Modes - **default** — the contract above. - **`deep <topic>`** — caps ×2.5, and one section diagram is allowed (type-matched per `references/diagram-convention.md`). Every gap gets its own evidence trail. Use only when the user asks for depth. - The user wants something they can click through step by step → hand off to `problem-description`. - The question is really "how *should* this work / what should we do" → hand off to `study-design-verify`. If the answer is requested as a **Markdown file**, it becomes a generated document: Rule 1 of the diagram convention applies (one overview Mermaid diagram at the top). ## Worked example > **/feynman Dataverse plugin pipeline** ``` 📘 Dataverse plugin pipeline ① อธิบายง่าย ๆ โค้ดที่เราแทรกเข้าไปให้ทำงานตอนมีคนบันทึกข้อมูล Dataverse จะหยุดกลางทาง เรียกโค้ดของเรา แล้วเดินต่อ เราเลือกได้ว่าให้แทรกก่อนหรือหลังเขียนลงฐานข้อมูล เหมือนสายพานตรวจของ — ของวิ่งผ่าน 4 จุด ใครจะแทรกก็เลือกจุดได้ ② จุดที่ยังคลุมเครือ | จุดที่พูดคลุม | คำถามจริงที่ค้าง | | "หยุดกลางทาง" | ถ้าโค้ดเรา throw แล้วข้อมูลที่เขียนไปแล้วย้อนกลับไหม | | jargon | อธิบายง่ายได้? | | stage | ได้ — จุดบนสายพาน | | async | ครึ่ง ๆ — ยังไม่ได้พูดว่าออกนอก transaction | ③ เติมช่องว่าง Gap 1 — throw แล้ว rollback ไหม → sync stage อยู่ใน transaction เดียวกัน ยกเว้น PreValidation และ async ที่อยู่นอก → พูดง่าย ๆ ว่า "throw ตอน sync = ยกเลิกทั้งชุด" (ที่มา: learn.microsoft.com/power-apps/developer/…/event-framework) ④ อธิบายใหม่ Dataverse เรียกโค้ดเราได้ 4 จุดรอบการบันทึก: PreValidation (นอก transaction), PreOperation (ก่อนเขียน), PostOperation (หลังเขียน) และ async (หลังจบ ไม่ร่วม transaction) สามจุดกลางอยู่ใน transaction เดียวกับการเขียน — โค้ดเรา throw คือยกเลิกทั้งชุด 1. เลือก stage = เลือกว่าเห็นข้อมูลตอนไหน 2. sync = ล้มพร้อมกัน, async = ล้มแยกกัน 3. อยากแก้ค่าก่อนบันทึก ต้อง PreOperation
From ThodsaphonSonthiphin/workflow-daily-work, MIT.
More from ThodsaphonSonthiphin/workflow-daily-work
- 0.529score
fit-gap-analysis
ThodsaphonSonthiphin
Compare a target (spec, product vision, competitor, RFP, or "to-be" design) against a system as actually built, and produce an evidence-first fit-gap — a…
SkillBFreshness: BClaude Code +9 - 0.529score
extract-findings
ThodsaphonSonthiphin
Normalize ANY input into findings.json — the first step toward turning review notes into an Azure DevOps backlog.
SkillBFreshness: BClaude Code +9 - 0.529score
ado-auth
ThodsaphonSonthiphin
Authenticate to Azure DevOps and fix auth problems — the shared primitive every other ado-* skill relies on.
SkillBFreshness: BClaude Code +9 - 0.529score
ado-create-work-items
ThodsaphonSonthiphin
Create Azure DevOps work items from a backlog_input.json (the output of classify-work-items) via the bundled create-backlog.cs script.
SkillBFreshness: BClaude Code +9 - 0.529score
ado-writeback-tracking
ThodsaphonSonthiphin
Add traceability columns to a spreadsheet source and write created Azure DevOps ticket IDs/URLs back into it, matched row-by-row by key.
SkillBFreshness: BClaude Code +9 - 0.529score
findings-to-ado-backlog
ThodsaphonSonthiphin
End-to-end orchestrator that turns findings from ANY input (audit spreadsheet, code/security review, QA report, meeting notes, a pasted list of issues) into…
SkillBFreshness: BClaude Code +9 - 0.529score
findings-to-github-issues
ThodsaphonSonthiphin
End-to-end orchestrator that turns findings from ANY input (audit spreadsheet, code/security review, QA report, meeting notes, a pasted list of issues) into a…
SkillBFreshness: BClaude Code +9 - 0.529score
document-what-shipped
ThodsaphonSonthiphin
Write and publish ONE documentation page about something that already ships - a user manual, a process and flow page, a release note, a runbook, or a rules…
SkillBFreshness: BClaude Code +9 - 0.529score
chart-map
ThodsaphonSonthiphin
Chart a Decision map for an effort too big for one agent session — name the destination, grill breadth-first to separate real decision tickets from fog…
SkillBFreshness: BClaude Code +9 - 0.529score
classify-github-issues
ThodsaphonSonthiphin
Map triaged findings to GitHub Issues with the correct labels, milestone, and body, and emit github_backlog_input.json ready for creation.
SkillBFreshness: BClaude Code +9 - 0.529score
classify-work-items
ThodsaphonSonthiphin
Map triaged findings to the correct Azure DevOps work item types for the TARGET project's process, and emit backlog_input.json ready for creation.
SkillBFreshness: BClaude Code +9 - 0.529score
copy-audit
ThodsaphonSonthiphin
Find every copy of a plugin or skill on this machine and prove which ones are stale.
SkillBFreshness: BClaude Code +9
Signals
- Stars
- 0
- Forks
- 0
- Open issues
- 6
- Last commit (this folder)
- Sep 5, 2026 (1 month ago)
From GitHub (fetched Oct 24, 2026).
Details
- License
- MIT
- First seen
- Oct 24, 2026
- Last checked
- Oct 24, 2026
- Source
- GitHubgithub-graphql
More from ThodsaphonSonthiphin
- 0.529score
extract-findings
ThodsaphonSonthiphin
Normalize ANY input into findings.json — the first step toward turning review notes into an Azure DevOps backlog.
SkillBFreshness: BClaude Code +9 - 0.529score
document-what-shipped
ThodsaphonSonthiphin
Write and publish ONE documentation page about something that already ships - a user manual, a process and flow page, a release note, a runbook, or a rules…
SkillBFreshness: BClaude Code +9 - 0.529score
ado-create-work-items
ThodsaphonSonthiphin
Create Azure DevOps work items from a backlog_input.json (the output of classify-work-items) via the bundled create-backlog.cs script.
SkillBFreshness: BClaude Code +9 - 0.529score
ado-writeback-tracking
ThodsaphonSonthiphin
Add traceability columns to a spreadsheet source and write created Azure DevOps ticket IDs/URLs back into it, matched row-by-row by key.
SkillBFreshness: BClaude Code +9 - 0.529score
ado-auth
ThodsaphonSonthiphin
Authenticate to Azure DevOps and fix auth problems — the shared primitive every other ado-* skill relies on.
SkillBFreshness: BClaude Code +9 - 0.529score
chart-map
ThodsaphonSonthiphin
Chart a Decision map for an effort too big for one agent session — name the destination, grill breadth-first to separate real decision tickets from fog…
SkillBFreshness: BClaude Code +9
Cite this page
feynman-explain skill. Agent Mart. https://agentmart.dev/skills/thodsaphonsonthiphin-feynman-explain (last updated 2026-10-24)