grilling: reword the primitive for general use - #532
Merged
Conversation
🦋 Changeset detectedLatest commit: 3bb587f The changes in this PR will be included in the next version bump. This PR includes changesets to release 1 package
Not sure what this means? Click here to learn what changesets are. Click here if you're a maintainer who wants to add another changeset to this PR |
mattpocock
force-pushed
the
grill-me-general-use
branch
2 times, most recently
from
July 13, 2026 09:58
60519e1 to
ce85654
Compare
mattpocock
changed the title
mattpocock
force-pushed
the
grill-me-general-use
branch
from
July 13, 2026 10:11
ce85654 to
c7ef452
Compare
Drop the software-plan scoping from grilling's description and body so the interview reads as a stress-test of any plan, decision, or idea: "this plan" → "this", "enact the plan" → "act on it", "codebase" → "environment". The technique is unchanged, and grill-me is left as-is. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
mattpocock
force-pushed
the
grill-me-general-use
branch
from
July 13, 2026 10:12
c7ef452 to
170ad48
Compare
Align the remaining callers and docs pages with grilling's body, which now says "decision tree". Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
This was referenced Jul 13, 2026
Merged
This was referenced Jul 13, 2026
kyle-sexton
added a commit
to melodic-software/claude-code-plugins
that referenced
this pull request
Jul 14, 2026
…sary discipline + domain-routing (#163) ## What Absorbs the current state of Matt Pocock's grilling-family skills (`mattpocock/skills`, ref [PR #532](mattpocock/skills#532)) into our `planning`/`improve-architecture` skills — the rewording **and** the generalization mindset behind it. Two separable commits. ### Commit 1 — reword + ADR/glossary discipline (`69df12e`) Safe, low-risk. Ready independently. - `design tree` → `decision tree` across `interview` me-mode + `improve-architecture` (finishes a rename already applied elsewhere; removes a pre-existing outlier). - Generalizes `interview` me-mode canonical framing to domain-neutral wording ("the task", "environment") to match the bare `grilling` primitive. Engineering *mechanics* keep "codebase" — the same primitive-vs-variant boundary Matt draws (`grill-with-docs` kept "codebase"). - Backs `interview`'s existing "ADRs are first-class" claim with an actual **discipline**: ADR offered sparingly via a 3-gate test (hard-to-reverse AND surprising AND real trade-off), plus a glossary **purity guard** + canonical-term/record-rejected-synonyms principle. Both **format-externalized** — discovery-first, consumer convention wins, no prescribed filename/location (per the extensibility contract + Rule of Three). ### Commit 2 — domain-route `interview` (`87b4c18`) — identity-level, closes #162 Makes `interview` general-purpose instead of engineering-welded — the headline of Matt's generalization. - The relentless-interview **loop is universal**; engineering machinery (codebase gate, PLAN.md Brief, ADR/glossary, `/architect` handoff) engages **only** when Step 1 Survey classifies the domain as engineering. - Domain is **inferred from context** (repo, cwd, the problem), never asked; orthogonal to `me`/`auto`/`lock`; user can override in prose. - **Output follows the domain**: engineering → PLAN.md Brief that feeds the pipeline; general → shared-understanding summary, terminal. **Handoff follows the artifact** — a general session offers no pipeline handoff, so the `/architect` contract is preserved without re-welding. - `description` broadened in *capability* but proactive auto-fire kept scoped to ambiguous behavior-changing work (verified against the skills doc this session) so it doesn't over-fire on casual decisions. ## Why two commits, one PR The reword (commit 1) is safe and could merge alone; the domain-routing (commit 2) is identity-level. Kept separable so commit 2 can be reviewed — or dropped — independently, while landing the whole absorption together. ## Deferred (tracked, not in this PR) - **#161** — standalone `domain-modeling` skill (active glossary discipline + multi-context `CONTEXT-MAP`). Deferred on Rule of Three: no skill yet *maintains* a glossary file as a first-class output. - **#162** — closed by this PR (commit 2); its "open implementation decisions" (general-mode artifact shape, `loop.md` detail alignment, optional escape-hatch action) remain as follow-ups. - **Domain storytelling** — separate modeling technique, logged for later. ## Verification - Three independent fresh-eyes reviews (correctness, upstream-fidelity, ownership-fit) on commit 1: no blockers; confirmed `decision tree` eradicated repo-wide and the ownership call sound. - Commit 2 coherence-checked (fixed a stale "both modes write the same Brief" line); engineering-only assertions consciously scoped by the new Domain-routed section. 🤖 Generated with [Claude Code](https://claude.com/claude-code) <!-- CURSOR_SUMMARY --> --- > [!NOTE] > **Low Risk** > Changes are documentation and skill-manifest text for Claude Code plugins; no runtime code or auth/data paths are modified. > > **Overview** > **Planning `interview` (0.7.0)** is reframed as a **domain-routed** skill: the relentless Q&A loop stays universal, but **engineering-only** machinery (PLAN.md Brief, glossary/ADR writes, `/architect` pipeline handoff) runs only when Step 1 classifies the task by **build surface**, not cwd alone. **General** sessions end with a shared-understanding summary (no Brief, no pipeline handoff). The skill **description** and **checklist**/`loop.md` are updated to match. > > **Terminology and discipline:** `design tree` → **`decision tree`** in `improve-architecture` (0.1.2) and interview `me` mode; canonical `me` framing uses **task** and **environment** instead of plan/codebase-only wording. **ADR** proposals get a 3-gate rule (hard to reverse, surprising, real trade-off); **glossary** updates are engineering-only with purity/canonical-term guidance, discovery-first (no prescribed paths). > > **improve-architecture** README and deepening Phase 3 use **decision tree** wording only (version bump). > > <sup>Reviewed by [Cursor Bugbot](https://cursor.com/bugbot) for commit 1923fdf. Bugbot is set up for automated code reviews on this repo. Configure [here](https://www.cursor.com/dashboard/bugbot).</sup> <!-- /CURSOR_SUMMARY --> --------- Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
This was referenced Jul 14, 2026
kosako
added a commit
to kosako/agent-tools
that referenced
this pull request
Jul 14, 2026
上流 (mattpocock/skills#532) の grilling 汎用化と突き合わせた残余対応: - 「コードベースで答えられる質問」を「調べれば分かること (コードベース・手元の資料・ 公開情報など)」に一般化 (企画・仮説の壁打ちにコードベースは無い) - 「合意前に動かない」を明文化 (共通理解の確認前に実装・実行・成果物作成に 着手しない)。description = 発火条件は不変。 grill-with-docs の同種の行は repo 密着 skill のため現状維持 (issue に記録)。 Claude-Session: https://claude.ai/code/session_01Wqz3c5ope2oVauyyWZNkyv Co-authored-by: kosako <kosako@users.noreply.github.com> Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
This was referenced Jul 14, 2026
This was referenced Jul 16, 2026
This was referenced Jul 27, 2026
This was referenced Aug 4, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Reword the
grillingprimitive so it reads as a stress-test of any plan, decision, or idea — not only a software build plan.grill-meis left untouched; the behaviour change lives in the shared primitive it delegates to.Changes
grilling/SKILL.md— description generalized ("a plan, decision, or idea … stress-test their thinking"); body drops the plan scoping:this plan→this,enact the plan→act on it,exploring the codebase→exploring the environment. The one-question-at-a-time discipline and the "design tree" concept are unchanged.The wrappers that invoke
/grilling(grill-with-docs,improve-codebase-architecture,triage,wayfinder) supply their own subject, so generalizing the wording doesn't affect them.🤖 Generated with Claude Code