aikey-concept: Retell a conceptThis skill was originally named
/zmm-…, and from v1.0.0 it is renamed AI KEY（/aikey-…), the old name is no longer updated. Existing users upgrading see README。
Phase 1, before locking the 3rd item, theory, read aikey/references/理论框架选型与装配.md §1–4, first compare explanations that are genuinely different, then score, verify and confirm according to the existing "lock four items". §5 mapping is merged into the one-pager without additional approval; the concept must still provide transferable judgment, and cannot replace definition with a philosophical label.
First read aikey/references/家族公约.md(if it cannot be read → explicitly say "the convention cannot be read, red lines cannot be guaranteed" and stop,do not substitute the convention from memory），
This skill's built-in criteria are in references/规则卡.md(criteria / why / how to check / strength); read it once before starting;{vault} when there is a corresponding rule file, the vault prevails and the rule card is the base.
Then read aikey/references/交互规范.md（🔴 reading once does not count as done: at the end, follow §4 three-piece set — Recap · Before/After · next step offers numbered options; when information is missing, follow §4 and usemultiple-choice questionsto ask,ask only one at a time; for cases where it does not apply see §5), then read {vault}/00-规则与索引/文案框架-总入口.md(the only navigation).
🔴 Its place on the shortest path:it replaces /aikey-script, not added alongside it
选题 → 【重讲概念型走这儿】→ 红线审核 → 发
/aikey-concept 代替 /aikey-script
⚠️ aikey routing records one field-tested finding: this system once had 7 pre-publish gates, and the result was 13 finished drafts, 0 published.
The number of gates itself is the cause of not publishing.
🔴 This skill does not add any gate. It does only one thing:move the rework that was going to happen anyway to happen earlier on one page.
Why this skill exists (measured cost, do not delete)
N22 "The lower the price, the worse it sells" went through 12 versions. Of the 12 versions, only two actually mattered: ① The Australian AI godfather said "my experience cannot be the subject, Douyin's mass audience cannot reach it」 ② he spoke himself。
The other nine versions were guessing what he wanted, and the reason each guess was wrong was the same: the whole draft was written first, and only then was it discovered that the proposition level was wrong.
| Version | Where it died |
|---|---|
| v1 · v2 · v4 | Ontology definition not locked, patched with "his experience" |
| v3 | Theory not locked(Weber explains "it dropped and no one felt it", not "it dropped and no one bought") |
| v5 · v6 | Proposition not locked(shifted from "teach one point" to "teach the whole thing") |
⇒ six versions of rework, all died on the same four things. And each of those four things takes only one sentence.
🟢 N23 "There is no precise traffic" verified it works — but pushed the problem downstream (2026-08-29)
N23 was basically usable when retold as v1, iterated only to v4, and not once was there rework at the proposition level. Locking four things holds.
🔴 But everything this one lost was lost in Phase 3–5:
| What was lost | Which step it died at | Cost |
|---|---|---|
| The line spoken on the audience's behalf(the sole source of the urge to share) | Phase 4 — I wrote it into "suggest adding it back", he approved it,I only applied the other spot, this line never made it into the body | Not in the final cut. The share criterion for that itemcan't be read out |
| The landing point changed without being treated as a rewrite | After Phase 3 — capability knife → classification knife, and the landing-point sentence is "the script's ID card" | frontmatter / title / filename still carried the old landing point for a while |
| Rejected the right opening on a wrong speech-rate parameter | Phase 2 — used the ÷5.1 point value to compute the first-5-second budget, judged "can't fit alongside high-ticket" | He added it back in the final cut,it fits. Measured 6.88, parameter off by 35% |
⇒ Once the upstream is locked, rework doesn't disappear; it moves downstream.
The four spots in this update (Phase 2 speech rate · landing-point check after Phase 3 · landing into body text in Phase 4 · cover and save-to-disk in Phase 5) all come from N23, all are downstream.
Who it's spoken to
A solo knowledge creator who researches and edits alone. Three hard constraints: don't assume he has a team · no upward reporting (give judgments directly) · zero jargon(theory is still used, but terms never appear in the final draft).
🔴 The cross-cutting layer is always there：CLAUDE.md Floor rule 7 — After finishing each paragraph, reread it as "I" and write out the inner line he'd have at that moment. If you can't write it, the paragraph isn't done. No tone of blame or preaching anywhere in the piece.
Process
Phase 0 接题 + 两道准入闸门
Phase 1 🔴 锁四样（一页纸，不写整稿） ← 本技能的全部价值
Phase 2 写整稿（一次写完，不逐段确认）
Phase 3 🔴 交给他口述（D1）
Phase 4 只剪不扩 + 红线全表 + 差异表 + 规则入库
Phase 5 /aikey-title 归口 + 封面 + 落盘 + 建取数日历
Phase 0 · Take the topic + admission (no pointless questions)
The topic is taken from the topic pool in the directory where {config.paths.topic_pipeline} is located, or newly given by the user.
Two gates, skip either and don't proceed(criteria in the pool §2.3, go read it, don't rely on memory):
| # | Gate | One sentence |
|---|---|---|
| ① | Anchor point | Before he hears your new formulation,is there a word he's already familiar with that you took apart first？ |
| ② | Ruler criterion | What you hand over —can he use it to measure something you never mentioned? Can't measure it = it's still just a definition |
Also run the deduplication gate once（总入口 §2 layer 1): core conclusion × main analogy comparison + write out which one the sibling item is。
Phase 1 · 🔴 Lock the four (the core of this skill, see references/锁四样.md）
Ask it all at once, one page, don't write the full draft.
| # | What to lock | Who provides |
|---|---|---|
| 1 | Specific question | AI proposes candidates,he decides |
| 2 | 🔴 Ontology definition / the counterintuitive line | Only he can provide it |
| 3 | Classic theory + disciplinary common-sense target | AI finds and scores on five dimensions,he confirms |
| 4 | 2–3 plain-language analogies | AI proposes,he swaps |
🔴 If the four aren't all locked, don't write the full draft.
The theory of item 3: one card per entry; any entry missing the line 'how far was it verified' may not go into the final draft:
| Line | What to write |
|---|---|
| What problem it solves | One sentence |
| Mechanism | One causal sentence |
| Source | Name · Work · Year · Chapter or page (write to whatever level of detail you can reach) |
| How far was it verified | 读过原文 / 权威二手（教材、综述，写明是哪本） / 没核到 |
| Verbatim or paraphrase | Is the sentence quoted in the final draft a verbatim translation or a condensed paraphrase; for a paraphrase, say 'the gist is' in the draft |
Rules: if the same source cannot be verified twice in a row, write 没核到 and stop; do not expand the search further, and do not name it in the final draft; for a quotation seen through secondary literature, write 'cited in' and do not pretend to have read the original; when a name first appears, write 'country · identity · Chinese name (common foreign name)'.Naming a book in the final draft means handing the audience a verifiable source, and the audience really will check. 'I remember he said' and 'I have read the original' must look different on this card.
⚠️ Item 2 is the easiest to skip, and it is exactly the root of the six revisions. AI cannot write a sentence like 'pricing itself is part of the product' — it can only write functional descriptions ('pricing answers three questions'). If you cannot get item 2, stop here and wait; do not substitute theory or personal experience.
Phase 2 · Write the full draft (in one pass)
Follow the {vault}/04-模板/概念纠偏型口播稿模板.md six sections.
Within section ⑤, follow the aikey-script/references/概念型口播.md seven beats —— 🔴 must open and read; two beats are most easily missed：
② Say the audience's answer for them('the beat most easily missed, decides success or failure') · ⑥ Second scenario from another industry('all the sense of gain comes from this beat').
Report three things as soon as the draft is out:Duration range · Gap distribution table(count the seconds) · Full-table red-line scan result。
🔴 Give only a range for duration, not a point value（2026-08-29 总入口 §3 #2 re-cut):
原速区间 = 稿面汉字 ÷ 6.9 ~ ÷ 5.1
🔴🔴 2026-08-31 correction: the basis for the paragraph above was wrong; do not follow the old version. Measurement confirms the final cut used 1.2× speed ⇒ 6.88 is the final-cut rate, not the speaking rate。 Original speed = 1583 ÷ (230×1.2) = 5.74 characters/second.
⇒ The reason given on 08-29 when overturning the old ruling was 'N19 5.1 vs N23 6.88,35% spread」—— after removing the speed multiplier it is 5.1 vs 5.74, a 12.5% spread. That reason no longer holds. 🔴 The original-speed speaking rates of the three concept-type pieces are actually fairly close together：N19 5.1 · N20 5.32 · N23 5.74. ⇒ Closed:calculate original-speed speaking rate using the 5.3–5.8 characters/second range(multiple measurements); for the first-5-second budget, use the lower bound of the range to calculate capacity and the upper bound to calculate duration; the final cut is the accelerated number, so divide by the speed multiplier before entering it into the parameter table.
🔴 Draft-delivery action (write the four frontmatter fields together; missing one does not count as backfilled)：
稿面字数 / 成片秒数 / 倍速系数 / 原速语速 = 字数 ÷ (成片秒数 × 倍速)
⚠️ Do not swap the two rates: to calculate script length useoriginal speed; to calculate how many characters fit in the first 2/3/5 seconds usefinal-cut speed
(Douyin's 2s 跳出 / 5s 完播 / 观看趋势 are all on the final-cut timeline).
🔴🔴 Before using duration to reject a sentence, ask one question first: would a 35% error flip this conclusion? N23 measured cost: I calculated based on ÷5.1 that only 21 characters fit in the first 5 seconds, and on that basis rejected the "and high-ticket customers" he wanted to add. He added it back in the final cut, and it fit. —— When a parameter is uncertain, don't use it as a veto.
🔴 When handing it over, you must state clearly: this version is a draft for you to narrate, not a final script.
Phase 3 · 🔴 Hand it to him to narrate
No retelling, no shooting. Narrated back → archive as-is → then process.
🔴 First thing after the narration comes back: check whether the landing point changed (added in N23)
It's not about checking the wording, it's about checking the landing point. 口播稿输出格式：The landing-point sentence is this script's ID card — changing the landing-point sentence means changing the script, not revising it.
Ask one question: the landing point he said, and the one you locked in,are they the same sentence?
N23 measured: no, and his is better.
| AI script | His retelling | |
|---|---|---|
| Landing point | When not running paid traffic, you don't have the action of "aiming" | You can have precision marketing, but you shouldn't have precision traffic |
| What kind of knife it is | Capability knife(you can't do it) | ★ Classification knife(that thing is called something else) |
⇒ Landing point changed = change four things on the spot: frontmatter landing-point sentence · work title · cover large text · file name。 🔴 Don't drag it until right before publishing — N23 was still hanging on the old landing point for a while.
⚠️ When the landing point changes, it's usually for the better(he is reasoning, the AI is describing).This step is to capture it, not to correct him.
Phase 4 · Cut only, don't expand
Optimization can only cut from the retelling script, and cannot expand upward from any AI script.
Four must-dos：
- Full red-line table（
违禁词避雷表10 categories + absolutes + symbol level) — 🔴 not grep four words - Unify person(when explaining mechanisms, use generic terms consistently throughout; only switch when discussing a specific industry scenario, and switch only that paragraph)
- Internal consistency（
/aikey-reviewQuestion 6 (skeleton and breakpoints): do all numbers/times/facts in the text match up front to back) - Difference table + rules into the repository（
总入口§5 four steps: archive the original words → determine the attribution layer → write executable criteria → attach measured cost)
🔴🔴 The fifth item (added in N23, because I missed it):Anything approved to be added back must be put into the main text on the spot and verified with grep
Because of the rule "cannot expand upward from an AI script," the things "he deleted" in the difference tablecan only be listed and asked about, and you are not allowed to add them back yourself. But after he approves? — the original process breaks off right here.
N23 measured cost: the sentence spoken on behalf of the audience (theonlysource of sharing impulse), after he approved adding it back, I only landed another spot; this sentence never entered the main text ⇒ not in the finished cut ⇒ that criterion cannot be read this time.
🔴 Three steps, none can be skipped：
① 他批准 → ② 当场落进正文（不是写进"建议"栏） → ③ grep 原句验证在不在
⚠️ The same thing has already been taught once: 📣 The checklist quotes the original sentence, not the line number, becauseline numbers will silently point to a different sentence。 This item is of the same kind: a sentence written in the suggestions columnwill silently not exist。 Criterion: after revising, grep that original sentence. If grep finds nothing = it is not there.
Phase 5 · Title cover + visuals + saving to disk
Title covergoes through /aikey-title routing(must mark structural numbers S1–S12, and must answer "according to this title, does the opening of the body need to change").
Final draft saved to {vault}/_草稿/NN_认知_{标题}.md, format according to 00-规则与索引/口播稿输出格式.md。
The filename and the draft's save location must match —— if the save location changes, rename it; don't drag it out until right before publishing.
🔴 Where are the cover assets (N23 addition: I once looked in the wrong directory)
docs/brand/design-system.md ← 视觉系统（暖白纸底 · 档案绿 · 宋体黑 · 刻度线母题）
docs/brand/douyin/cover-template.html ← 抖音封面模板，改帧和标题即可
docs/brand/douyin/cover-00X.* ← 已发的封面，看它们对齐风格
⚠️ not in _装修/。 N23 I only searched _装修 and _已发布，did not search docs/,
and as a result made a white-background black-text pure-text cover, detached from the whole visual system, scrapped and redone.
The template's two hard rules(written in the template comments, go read them):
- Cover large text ≤6 characters per line, at most two lines（
口播稿输出格式2026-08-29 wording corrected: originally wrote "≤6 characters" and omitted "per line") - When both lines are 6 characters, the font size must be reduced(long-sentence mode 108/148), otherwise the size contrast is gone and the template falls apart
🔴🔴 The pitfall of grabbing a cover frame: subtitles are burned into the picture
The template's title area sits right on the subtitle band. N23 measured: the subtitle band is at y1400–1520 (75% of the frame).
Do not just grab any frame. Sample the whole video at 1.5-second intervals → programmatically filter for clean subtitle bands → then sort by contrast in the header area. N23 measured: out of 143 frames, only 4 had no subtitles.
🔴 After publishing (N23 addition)
| What to do | Why |
|---|---|
Register _已发布/已发布内容库.md —— core conclusion + main analogyFill in both keys |
If one is missing, next time duplicates cannot be judged. ⚠️ N20 was published and never registered; the deduplication gate is blind to it |
| Create a T+24h / T+7d data-collection calendar | T+24h is a snapshot, all accounts collect in the same window; reading late makes cross-comparison impossible |
| Hard-code in the calendar descriptionthe data-collection checklist + minimum play-count threshold | Not reaching the threshold = do not read this time,not that a negative result was read out |
| Copy "what this item cannot be attributed to" into the T+7d entry | To prevent skipping over it during review |
🔴🔴 Open the next draft's frontmatter 承接上条复盘: column, copy item by item this review's "N things that carry over to N+1", mark each as ✅ 已执行 / ❌ 已否决（理由） |
Measured: N20's review fixed "remove the 1.15× speedup in the next one"——N23 not only did not remove it, but increased it to 1.2×. The instruction was not rejected; it was silently lost.Cost: the editing layer and the copy layer have never once been separated to this day; that 24pp retention collapse on N23 could have been a clean read.Pre-publish gate: this field must not have blank entries. Vetoing is fine, silent loss is not. |
🔴 Six "must nots" (all have measured costs, not a single one may be softened)
| # | Must not | Cost |
|---|---|---|
| 1 | Write the full draft before the four things are locked | N22 v1–v6 Six rounds of rework |
| 2 | Use "my experience" as the skeleton | Three criteria: ① If you delete the entire "I" passage, does the argument still hold? ② If it takes up >1/4 of the body, it is the main subject ③ 🔴 Does the Douyin mass audience have this scenario? If not = it can only serve as one supporting sentenceN22 only passed after being compressed from 31% to 5% |
| 3 | Only grep the four prohibited words | 「USD" slipped all the way into the example for 作者方法论 §5.2 —— Once a prohibited word enters a long-term file, it will copy itself in every subsequent session |
| 4 | Cite rules from memory | I once cited 语言风格档案 without having read that file (I guessed right that time).The §6 loading list says that to write the body you must read it, so you really have to read it |
| 5 | Reconcile conflicting rules yourself | Check 总入口 §3 ruling table;List anything not in the table and ask; after a ruling, write it back into the table(this skill has already produced #13/#14/#15) |
| 6 | He makes the call to keep it, then adds it back himself after deletion | 裁定 #12: execute, and write the exemption next to it |
Load-on-demand list (save tokens, do not read it all at once)
| What I am doing now | Read at minimum these |
|---|---|
| Judge whether this concept can be done | This skill's Phase 0 + topic pool + 已发布内容库 |
| Lock the four things | references/锁四样.md(the only must-read file that comes with this skill) |
| Write the body | 04-模板/概念纠偏型口播稿模板.md + aikey-script/references/概念型口播.md + 语言风格档案 |
| Pair with theory | Three admission criteria + five-dimension scoring + theory card "to what level is it verified" (general disciplinary knowledge does not follow citation discipline） |
| Red-line self-check | 规则总表 condensed checklist + 违禁词避雷表 full table |
| Review | /aikey-review four decisions and eleven questions (if the vault has 发布前审核流程.md , that takes precedence) |
Absolutely never do
- Do not write the full draft before the four things are locked.
- Do not draw the conclusion "this is a misconception" for the audience —— you only lay out the crack; that sentence is for them to say themselves.
- Do not do conversion actions —— no "comment 1 / DM / link in the comments / contact me if you need it" of any kind.
- Do not use examples the audience cannot reach (for luxury examples, look atwhether the audience stands in the seller's or the buyer's position）。
- Do not force in a book just to have an endorsement (
§1.65if mechanism match <2, eliminate directly). - Do not add new gates.
Don't know the next step → go back to /aikey。