資料來源#
摘要#
Matt Pocock 的 grill-me skill 是一種步步緊逼的訪談提示,會沿著決策樹逐一追問,每個問題都附上建議答案。它把「請代理程式擬定計畫」改成「在任何計畫成形前先取得共同理解」。重點是對齊,而非產出。目標狀態就是 Frederick P. Brooks 在《The Design of Design》中所稱的設計概念:所有參與工作的人共同持有的一個想法。PRD 或計畫都位於設計概念之後;尚未先取得共識就產出文件,必然會造成返工。
這項 skill(原文)#
"Interview me relentlessly about every aspect of this plan until we reach a shared understanding. Walk down each branch of the decision tree, resolving dependencies one by one. For each question provide your recommended answer. Ask the questions one at a time…"
就是這樣。這項 skill 刻意保持簡短——盡可能縮小表面範圍,盡可能大幅改變行為。
為什麼要先盤問再規劃#
Pocock 觀察到,代理程式在計畫模式下「真的會急著產出計畫」——只要說「我想我已經掌握得夠多了」,就會交出一份掩蓋未解問題的計畫。計畫讀起來沒問題,實際上卻有錯,直到開始實作才會浮現。強迫代理程式先訪談,能在還來得及以低成本回答時,先把未解問題挖出來。
每個問題都附上建議答案,這種模式是關鍵:大多數時候,使用者只要說「好,同意」,只有在確實意見不同時才需要討論。只提問題的訪談,會把使用者的注意力浪費在顯而易見的決定上。
對「規格轉程式碼」運動的反向主張#
Pocock 最鮮明的反面主張是:**規格轉程式碼只是換個名字的氛圍式編碼。**支持者會說:「寫一份仔細的規格交給 AI,程式碼錯了就修規格,永遠不用看程式碼。」Pocock 試過了:行不通。
原因:
- 真正的戰場是程式碼,不是規格
- 不接觸程式碼的規格會退化成願望清單
- 回饋迴路經過一層(規格 ⇄ AI ⇄ 程式碼),而不是直接在錯誤真正所在之處運作(程式碼 ⇄ 測試)
- 不接觸程式碼,開發者對系統的心智模型就會腐朽
Grilling 採取相反的紀律:規格是在對齊之後才產生;對齊先於任何產物;開發者從頭到尾都親自參與程式碼工作。
Grilling 會產生什麼#
一場 grilling 訪談可能包含 10 到 100 個問題;Pocock 有些訪談進行了整整一小時。最後留下的產物是對話紀錄本身——保留為 PRD 階段的原始素材。Pocock 的 write-a-PRD skill 會使用這份紀錄(以及另一段簡短訪談)來產出目的地文件。
他明確表示之後不會審閱 PRD:
"What am I testing at this point? What are the failure modes I'm trying to test for? I know that LLMs are great at summarization. I have reached the same wavelength as the LLM. So all I'm doing is checking the LLM's ability to summarize."
只有在 grilling 訪談已完成對齊工作時,這樣做才安全。跳過 grilling,就必須閱讀 PRD。
兩份必要文件#
Grilling 之後,Pocock 只會產出兩份文件:
- PRD(目的地文件)——完成後的成品樣貌、使用者故事、完成定義、範圍外清單、實作決策、測試決策,以及要修改的模組
- Kanban(旅程文件)——拆成可各自領取的垂直切片票單(參見 Vertical Slice Tracer Bullets)
實作完成後,他會刪除(或關閉)PRD——參見 doc rot。
PRD 中會列出模組地圖#
PRD 會列出「要修改的模組」——具體指出哪些現有模組會變更,以及會新增哪些模組。這讓規劃與架構連結起來(參見 Deep Modules for Agents)。重點是在整個規劃過程中都把程式碼庫的形狀放在心上,而不是等到實作時才想到。
何時可以略過 grilling#
Grilling 適用於有人在迴圈中的任務。若只是範圍明確的短小變更(「在整個程式碼庫中重新命名這個函式」),花在 grilling 上的成本就浪費了。這套紀律會隨利害程度調整:功能越大、需求越模糊、走錯方向的代價越高 → 就越該深入盤問。
相關連結#
-
Spec-Driven Development as the New Waterfall — 這是此模式必須能承受的反對意見:Martin 的目標是讓代理程式逐步執行一份可操作的計畫,而不是 grilling 所產生的共同理解產物
-
Matt Pocock — 這項 skill 的作者
-
Vertical Slice Tracer Bullets — PRD 之後接續的 Kanban
-
Deep Modules for Agents — PRD 中的模組地圖將規劃連結至架構
-
Agent Loop Pattern — grilling 位於漏斗頂端、有人參與的部分;迴圈則負責漏斗底端的 AFK 工作
-
Context Window Smart Zone — grilling 使用子代理程式來縮小父層的上下文
-
Agent Harness Engineering — 在規劃層「強制執行不變條件」,就是「在任何計畫出現前先取得共識」
-
Claude Code Best Practices — explore→plan→code 工作流程的形狀相同;grill-me 是其中「explore」步驟更積極的變體
-
Interaction Models — grilling 本身就是即時協作迭代;turn-based interfaces 正是它如今顯得笨拙的原因,而互動模型則是讓 grilling 式協作感覺自然的基礎
-
HTML as the New Markdown — 腦力激盪 → 讓 Claude 訪談你 → 規劃,就是 grilling 的形狀;Thariq 的 HTML 計畫是比 Markdown PRD 更豐富的目的地產物,但版本管理也更困難
-
Agentic Technical Debt — grilling 產生的設計概念會寫入 CLAUDE.md;這是防止每次工作階段重新推導而累積技術債的最強上游防線
-
Zero-Friction Scope Creep — 透過 grilling 取得有力的設計概念,能以書面 PRD 常做不到的方式抵擋範圍膨脹
-
Evals as Product Spec — grilling 產生設計概念;evals 則記錄它是否達成。Matt 所說的「verification loops」與 Cat 所說的「ten great evals」,是分別位於規劃另一端的同一種基本做法
-
Building Is Cheap, Arguing Is Expensive — 一種富有成效的張力:Fiona Fung 所說的「產生三個 PR 再比較」,是把設計移至已建成的產物中;在此可以調和為:原型是呈現設計概念的媒介,而不是取代達成設計概念的方式
尚待釐清的問題#
- 能不能讓 grilling 在 AFK 狀態下對另一個掌握使用者偏好的代理程式執行?Pocock 在 2026 年的回答是「不行,這部分必須有人參與」——但隨著代理程式越來越擅長建模其委託人,這個問題仍有待探討。
- 多位人類需要取得共識的團隊工作中,grilling 會如何改變?Pocock 的提示是:讓代理程式加入現場的結對程式設計,把它當成第三位對話者。
衍生文章#
- The PRD-Replacement Spectrum at AI-Native Speed — 光譜的左端:建置前達到最大程度的對齊,PRD 作為事後刪除的目的地文件
- Where Does the Why Live? — grilling 訪談是撰寫當下安放「為什麼」的所在;但承載這個理由的目的地 PRD 會被刪除,因此後續讀者無從得知
- Is Persistence the Line Between Prompting and Spec-Driven Development? — 本文是打破「持久性即定義」的納入不足案例:整個語料中立場最偏重規格的文章,仍以短暫對話作為產物,而 Pocock 自己的 PRD 其實是關閉而非刪除——撤銷其效力,但保留檔案
資料來源#
Cited by 32
- The PRD-Replacement Spectrum at AI-Native Speed×7
Design Concept Grilling — the left pole: maximal pre-build alignment, PRD as deleted destination doc
- Where Does the Why Live?×6
The destination PRD · Held the why momentarily, then deleted after implementation — so the why dies…
- Learning to Co-Work with AI: A Software Engineer's Field Guide×5
What it is: reaching shared understanding (the "design concept" in Frederick Brooks's sense) before…
- Human-in-the-Loop Boundaries×4
Design Concept Grilling identifies the upstream loop that should remain human-heavy: before a plan,…
- Prototype Over PRD×4
Dan Carey's account of how the Claude Design team replaced product requirements documents with…
- Spec-Driven Development as the New Waterfall×4
Code As Source Of Truth and Committed Artifact Chain hold the opposite practice on the point he is…
- Is Persistence the Line Between Prompting and Spec-Driven Development?×3
The grilling transcript. Design Concept Grilling is the left pole of the PRD-replacement spectrum —…
- Unknowns as the Agentic Bottleneck×3
The wiki's existing planning discipline — grill first, then plan — assumes unknowns can be drained…
- Agent Loop Pattern×2
Amortize planning over many executions. One careful planning session (e.g. via Design Concept…
- Opinions on Using AI Tools & the Future of the Software Engineering Role×2
Specs-to-code is "vibe coding by another name." The code is the battleground, not the spec; reach a…
- Building Is Cheap, Arguing Is Expensive×2
This norm is in productive tension with the wiki's planning-first concepts. Design Concept Grilling…
- Claude Code Best Practices×2
A more aggressive variant: Design Concept Grilling (Matt Pocock's grill-me skill) replaces "ask the…
- Deep Modules for Agents×2
Design Concept Grilling — module map in the PRD operationalizes deep-module discipline at planning…
- Evals as Product Spec×2
Design Concept Grilling / Deep Modules For Agents — Matt Pocock's verification-loop framing; same…
- Matt Pocock×2
Grill, then plan, not plan, then read. Uses a relentless-interview skill (grill-me) to reach…
- Open Questions Backlog×2
Design Concept Grilling (152d) — How does grilling change for team work where multiple humans need…
- Playbook Boundary Conditions: the Devil's-Advocate Substrate and the Prototype's Edge×2
Cross-team coordination — the artifact aligns the people in the room with it (the pitch-off); a…
- Agent Harness Engineering
Design Concept Grilling — alignment-layer harness primitive; prevents premature plan generation by…
- Agentic Technical Debt
Design Concept Grilling — Matt Pocock's grill-me pattern reaches a Brooks "design concept" before…
- Configurable Human Participation
Design Concept Grilling — agent-initiated clarification scored by the Clarification Quality Score…
- Context Window Smart Zone
Design Concept Grilling — the grilling session uses a sub-agent so the parent context stays small
- HTML as the New Markdown
Design Concept Grilling — brainstorm → let Claude interview you → plan is the grilling shape; the…
- Interaction Models
Design Concept Grilling — collaborative, real-time iteration over spec-and-walk-away; an…
- LLM-as-Compiler Knowledge Base
Design Concept Grilling — Brooks's "design concept" (shared understanding before any artifact) is…
- AI Coding Practice
Design Concept Grilling (hub) — Matt Pocock's grill-me skill; reach Brooks "design concept" before…
- Model Introspection Feedback
Design Concept Grilling — the inverse: model interviews user. Same pattern, opposite direction.…
- Model Spec Midtraining (MSM)
Product spec (Symphony, Design Concept Grilling): humans + agents read the spec to align on…
- Outsource Your Thinking, Not Your Understanding
Design Concept Grilling — reaching the Brooks "design concept" before planning is forcing…
- Rationale as a Dated Record: Where the Why Lives for the Next Reader
The closed issue · the PRD kept after implementation with its authority revoked · One…
- Turn-Based Interface Bottleneck
Design Concept Grilling — argues the value is in collaborative iteration; this page argues the…
- Vertical Slice Tracer Bullets
Design Concept Grilling — the slice list comes out of the PRD that comes out of the grilling session
- Zero-Friction Scope Creep
Design Concept Grilling — Matt Pocock's grill-me pattern as upstream defense (a strong design…
Related articles
- Harness Shrinkage as Models Improve
Prompt scaffolding shrinks each model release; Cat Wu's pruning discipline; Boris Cherny "100 lines of code a year from…
- Deep Modules for Agents
Ousterhout deep-vs-shallow modules applied to agent-friendly codebases; push-vs-pull instruction delivery; reviewer in…
- Context Window Smart Zone
Smart zone vs dumb zone (Dex Horthy / Matt Pocock): quadratic attention scaling, ~100K marker independent of advertised…
- Matt Pocock
Independent AI-coding educator; built Sandcastle library; smart-zone/grill-me/tracer-bullets pedagogical framing; "bad…
- Claude Code
Anthropic's agentic coding product; created by Boris Cherny late 2024; TypeScript/React on Bun (itself Claude-rewritten…
