@pmelab/gtd 51.0.1 → 52.1.0

package/src/workflows/feature/slicing.feature

@@ -36,6 +36,10 @@ Feature: a change with more criteria than the cap is built in coarse slices
       """
       # Requirements
 
+      ## Impact
+
+      - the greeting module
+
       ## Acceptance Criteria
 
       - AC-1: subtract(a, b) returns a - b.
@@ -47,10 +51,6 @@ Feature: a change with more criteria than the cap is built in coarse slices
       it("[AC-2] multiplies", () => {})
       """
     And step "seam.naive" is prompted with "Read `.gtd/REQUIREMENTS.md`"
-    And judge step "implement.coverage" answers:
-      """
-      [{"id":"AC-1","answer":"yes","p":0.9},{"id":"AC-2","answer":"yes","p":0.9}]
-      """
     When the process is driven
     Then the process rests at "build.review.reviewing"
     And the commit subjects end with:
@@ -59,8 +59,7 @@ Feature: a change with more criteria than the cap is built in coarse slices
       gtd(agent): seam.naive → seam.check
       gtd(check): seam.check → implement.applying
       gtd(check): implement.applying → implement.health.check
-      gtd(check): implement.health.check → implement.coverage
-      gtd(judge): implement.coverage → implement.closing
+      gtd(check): implement.health.check → implement.closing
       gtd(check): implement.closing → build.review.reviewing
       """
     And the output lines of gtd "slices" start with:
@@ -75,6 +74,10 @@ Feature: a change with more criteria than the cap is built in coarse slices
       """
       # Requirements
 
+      ## Impact
+
+      - the greeting module
+
       ## Acceptance Criteria
 
       - AC-1: subtract(a, b) returns a - b.
@@ -104,14 +107,6 @@ Feature: a change with more criteria than the cap is built in coarse slices
       """
       it("[AC-2] multiplies", () => {})
       """
-    And judge step "implement.coverage", when its questions contain "AC-3", answers:
-      """
-      [{"id":"AC-1","answer":"yes","p":0.9},{"id":"AC-3","answer":"yes","p":0.9}]
-      """
-    And judge step "implement.coverage", when its questions contain "AC-2", answers:
-      """
-      [{"id":"AC-2","answer":"yes","p":0.9}]
-      """
     When the process is driven
     Then the process rests at "build.review.reviewing"
     And the commit subjects end with:
@@ -122,14 +117,12 @@ Feature: a change with more criteria than the cap is built in coarse slices
       gtd(agent): seam.naive → seam.check
       gtd(check): seam.check → implement.applying
       gtd(check): implement.applying → implement.health.check
-      gtd(check): implement.health.check → implement.coverage
-      gtd(judge): implement.coverage → implement.closing
+      gtd(check): implement.health.check → implement.closing
       gtd(check): implement.closing → seam.naive
       gtd(agent): seam.naive → seam.check
       gtd(check): seam.check → implement.applying
       gtd(check): implement.applying → implement.health.check
-      gtd(check): implement.health.check → implement.coverage
-      gtd(judge): implement.coverage → implement.closing
+      gtd(check): implement.health.check → implement.closing
       gtd(check): implement.closing → build.review.reviewing
       """
     And the output lines of gtd "slices" start with:
@@ -148,6 +141,10 @@ Feature: a change with more criteria than the cap is built in coarse slices
       """
       # Requirements
 
+      ## Impact
+
+      - the greeting module
+
       ## Acceptance Criteria
 
       - AC-1: subtract(a, b) returns a - b.
@@ -171,6 +168,10 @@ Feature: a change with more criteria than the cap is built in coarse slices
       """
       # Requirements
 
+      ## Impact
+
+      - the greeting module
+
       ## Acceptance Criteria
 
       - AC-1: subtract(a, b) returns a - b.
@@ -196,6 +197,10 @@ Feature: a change with more criteria than the cap is built in coarse slices
       """
       # Requirements
 
+      ## Impact
+
+      - the greeting module
+
       ## Acceptance Criteria
 
       - AC-1: subtract(a, b) returns a - b.
@@ -206,6 +211,10 @@ Feature: a change with more criteria than the cap is built in coarse slices
       """
       # Requirements
 
+      ## Impact
+
+      - the greeting module
+
       ## Acceptance Criteria
 
       - AC-1: subtract(a, b) returns a - b.
@@ -230,6 +239,10 @@ Feature: a change with more criteria than the cap is built in coarse slices
       """
       # Requirements
 
+      ## Impact
+
+      - the greeting module
+
       ## Acceptance Criteria
 
       - AC-1: subtract(a, b) returns a - b.
@@ -253,16 +266,16 @@ Feature: a change with more criteria than the cap is built in coarse slices
       """
       it("[AC-1] subtracts", () => {})
       """
-    And judge step "implement.coverage" answers:
-      """
-      [{"id":"AC-1","answer":"yes","p":0.9}]
-      """
     When the process is driven
     Then the process rests at "seam.steer"
     Given a file ".gtd/REQUIREMENTS.md" with:
       """
       # Requirements
 
+      ## Impact
+
+      - the greeting module
+
       ## Acceptance Criteria
 
       - AC-1: subtract(a, b) returns a - b.
@@ -276,8 +289,7 @@ Feature: a change with more criteria than the cap is built in coarse slices
       gtd(agent): seam.naive → seam.check
       gtd(check): seam.check → implement.applying
       gtd(check): implement.applying → implement.health.check
-      gtd(check): implement.health.check → implement.coverage
-      gtd(judge): implement.coverage → implement.closing
+      gtd(check): implement.health.check → implement.closing
       gtd(check): implement.closing → build.review.reviewing
       """
     And ".gtd/REQUIREMENTS.md" does not exist

package/src/workflows/feature/standards.feature

@@ -37,6 +37,10 @@ Feature: the review learns standards from the human's answers
 
       - `subtract(a, b)` returns `a - b`.
 
+      ## Impact
+
+      - the greeting module
+
       ## Acceptance Criteria
 
       - AC-1: subtract(a, b) returns a - b.
@@ -55,10 +59,6 @@ Feature: the review learns standards from the human's answers
     And step "implement.applying" runs its script
     And step "implement.health.check" changes nothing
     And step "implement.closing" runs its script
-    And judge step "implement.coverage" answers:
-      """
-      [{"id":"AC-1","answer":"yes","p":0.9}]
-      """
     And step "build.quality.correctness.reviewing" writes ".gtd/QUALITY.md" with:
       """
       # Quality

package/src/workflows/feature/tier-pin.feature

@@ -34,6 +34,10 @@ Feature: a process keeps the risk tier setting it started with when GTD_FLAG_RIS
 
       low — a new pure function
 
+      ## Impact
+
+      - the greeting module
+
       ## Acceptance Criteria
 
       - AC-1: the change behaves as described.

package/src/workflows/shared/review.ts

@@ -16,7 +16,14 @@ import {
   type Judgment,
   type ScopeAccess,
 } from "../../flows/index.js"
-import { reviewChunks, reviewRisks, type ReviewNote } from "../../steering/index.js"
+import {
+  REVIEW_SECTIONS,
+  reviewChunks,
+  reviewRisks,
+  reviewSections,
+  sectionedReview,
+  type ReviewNote,
+} from "../../steering/index.js"
 import * as kit from "../../workflow-kit/index.js"
 import { guarded, healthy, type EscalationCount } from "./health.js"
 import { NO_FINDINGS, type LensReport } from "./lenses.js"
@@ -231,11 +238,16 @@ const batched = <T extends { readonly evidence: string }>(
 }
 
 /**
- * Rates every chunk the reviewer wrote — a walkthrough chunk or a carried
- * finding — and sorts \`full\` by it, each chunk led by its tier and reason.
- * gtd's summary chunks are not rated and stay last.
+ * Rates every chunk the reviewer wrote and sorts the \`findings\` among them
+ * first in \`full\`, each led by its tier and reason; a walkthrough chunk
+ * keeps its place and carries no tier. It is still rated, so a review with no
+ * findings still reaches \`review.ranking\` and older histories replay.
  */
-const ranked = async (review: string, full: string): Promise<string> => {
+const ranked = async (
+  review: string,
+  full: string,
+  findings: ReadonlySet<string>,
+): Promise<string> => {
   const texts = t.chunkTexts(full)
   const chunks = reviewChunks(review).map((c, i) => ({
     id: `C${i + 1}`,
@@ -268,6 +280,7 @@ const ranked = async (review: string, full: string): Promise<string> => {
     full,
     chunks
       .map((c, i) => ({ c, i, s: severities.get(c.id)! }))
+      .filter(({ c }) => findings.has(c.title))
       .sort((x, y) => chunkKey(x.s) - chunkKey(y.s) || x.i - y.i)
       .map(({ c, s }) => ({ title: c.title, label: s.reason })),
   )
@@ -342,6 +355,10 @@ const requirePointers = (
     ...chunks
       .filter((c, i) => chunks.findIndex((d) => d.title === c.title) !== i)
       .map((c) => `  - ${REVIEW}:${c.line + 1}: ${c.title} — a second chunk with this title`),
+    ...reviewSections(review).map(
+      (s) =>
+        `  - ${REVIEW}:${s.line + 1}: ${REVIEW_SECTIONS[s.section]} — gtd sorts the chunks into sections; write \`## \` chunks only`,
+    ),
     // A risk is fixed before the human sees it; a note is the human's to decide.
     ...reviewRisks(review)
       .filter((r) => [...noted].some((title) => r.anchor.startsWith(`${title} ./`)))
@@ -377,7 +394,10 @@ const walkthrough = async (
     changesSince(base).some((c) => isCode(c.path)),
   )
   // Ranked even when it marks risks: the review written after their fix reaches the human with its own.
-  const full = await ranked(review, t.withWalkthrough(review, report))
+  const full = sectionedReview(
+    await ranked(review, t.withWalkthrough(review, report), new Set(notes.map((n) => n.title))),
+    new Set(reviewChunks(review).map((c) => c.title)),
+  )
   if (full !== written) {
     await run("review.carrying", writeScript(REVIEW, full), {
       label: "Ordering the review and adding the quality lap's findings",

package/src/workflows/shared/text.test.ts

@@ -4,6 +4,8 @@ import { tmpdir } from "node:os"
 import { join } from "node:path"
 import { afterEach, describe, expect, it } from "vitest"
 import {
+  designImpactPrompt,
+  prototypeFixingPrompt,
   buildReviewTriageMessage,
   buildReviewReviewingPrompt,
   agentWithSkills,
@@ -19,9 +21,12 @@ import {
   buildReviewCollectingPrompt,
   designGateAnswerMessage,
   designTriagePrompt,
+  normaliseCriteriaScript,
+  prototypeTestingPrompt,
   implementBuildingPrompt,
   implementFixPrompt,
   naivePassPrompt,
+  slicingPrompt,
   refactorBuildingPrompt,
   refactorDroppedRefusal,
   refactorFixPrompt,
@@ -170,6 +175,37 @@ describe("the risk tier prompt and gate line", () => {
   })
 })
 
+describe("the (no hands-on) tag", () => {
+  it("is explained at triage and at the design gate", () => {
+    expect(renderText(() => designTriagePrompt("base"))).toContain(
+      "`- AC-<n> (no hands-on): <criterion>`",
+    )
+    expect(renderText(() => designGateAnswerMessage())).toContain(
+      "add or remove `(no hands-on)` between an `AC-<n>` and its colon",
+    )
+  })
+
+  it("names the tagged criteria the testing turn skips, and only when there are any", () => {
+    expect(renderText(() => prototypeTestingPrompt(undefined, ["AC-2", "AC-3"], false))).toContain(
+      "- Skip AC-2, AC-3: no person can observe them",
+    )
+    expect(renderText(() => prototypeTestingPrompt(undefined, [], false))).not.toContain("Skip")
+  })
+
+  it("is kept when the stated criteria are recorded", () => {
+    const script = normaliseCriteriaScript({
+      path: "TODO.md",
+      commit: "abc",
+      criteria: [
+        { id: "AC-1", text: "a" },
+        { id: "AC-2", text: "b", noHandsOn: true },
+      ],
+    })
+    expect(script).toContain("'- AC-1: a'")
+    expect(script).toContain("'- AC-2 (no hands-on): b'")
+  })
+})
+
 describe("the shared open-question instruction", () => {
   it("leaves the option count to the agent, floored at two real options", () => {
     const prompt = renderText(() => designTriagePrompt("base"))
@@ -889,6 +925,29 @@ describe("buildReviewReviewingPrompt scenarios", () => {
   })
 })
 
+describe("Impact in prompts", () => {
+  it("[AC-4] the slicing prompt slices along the units Impact names", () => {
+    const prompt = renderText(() => slicingPrompt(8))
+    expect(prompt).toContain("## Impact")
+    expect(prompt).toMatch(/slice\s+along\s+the\s+units\s+it\s+names/)
+  })
+
+  it.each([
+    ["naive", () => naivePassPrompt()],
+    ["implement", () => implementBuildingPrompt(undefined)],
+    ["prototype fixing", () => prototypeFixingPrompt(undefined)],
+  ])("[AC-5] [AC-7] the %s prompt gives Impact as hints, not allowed files", (_n, prompt) => {
+    const p = renderText(prompt).replace(/\s+/g, " ")
+    expect(p).toContain("`## Impact`")
+    expect(p).toContain("not a list of allowed files")
+  })
+
+  it("the prototype testing prompt, which changes no code, carries no Impact hints", () => {
+    const p = renderText(() => prototypeTestingPrompt(undefined, [], false))
+    expect(p).not.toContain("not a list of allowed files")
+  })
+})
+
 describe("the seam phase's texts", () => {
   const rejection = { strays: ["src/flows/helpers.ts", "src/flows/runtime.ts"], p: 0.85 }
 
@@ -1281,7 +1340,7 @@ describe("composedReview", () => {
     })
 
     it("asks about each undecided standard in a note that blocks sign-off", () => {
-      expect(withConflict).toMatch(/^## Standard conflict [0-9a-f]{8}$/m)
+      expect(withConflict).toMatch(/^## Findings\n\n### Standard conflict [0-9a-f]{8}$/m)
       expect(withConflict).toContain("- [ ] keep the new one (drop the old)")
       expect(unansweredNotes(withConflict)).toHaveLength(1)
     })
@@ -1317,8 +1376,8 @@ describe("composedReview", () => {
     })
     expect(steeringFormatFor("review")!.validate(doc)).toEqual([])
     expect(doc).toContain("# Review: 0123456\n\n<!-- base: " + BASE + " -->")
-    expect(doc).toContain("## same\n\none")
-    expect(doc).toContain("## same (2)\n\n_The fixer left this unresolved._\n\ntwo")
+    expect(doc).toContain("## Findings\n\n### same\n\none")
+    expect(doc).toContain("### same (2)\n\n_The fixer left this unresolved._\n\ntwo")
     expect(doc).toContain("## Fixed automatically\n\n- f (fedcba9)")
     expect(doc).toContain("## Dropped as noise\n\n- n (p=0.8)")
   })
@@ -1340,12 +1399,13 @@ describe("composedReview", () => {
       fixed: [{ title: "f", commit: "fedcba9876543210" }],
       noise: [],
     })
-    const titles = doc.split("\n").filter((l) => l.startsWith("## "))
+    const titles = doc.split("\n").filter((l) => /^#{2,3} /.test(l))
     expect(titles).toEqual([
-      "## a",
-      "## a (2)",
-      "## a (3)",
-      "## Fixed automatically (2)",
+      "## Findings",
+      "### a",
+      "### a (2)",
+      "### a (3)",
+      "### Fixed automatically (2)",
       "## Fixed automatically",
     ])
   })
@@ -1376,10 +1436,10 @@ describe("composedReview", () => {
       noise: [],
     })
     expect(doc).toContain(
-      "## a\n\n- ends in a list\n\nTick exactly one:\n\n- [ ] fix as suggested\n- [ ] won't fix\n- [ ] _your answer_\n\n## Fixed automatically",
+      "### a\n\n- ends in a list\n\nTick exactly one:\n\n- [ ] fix as suggested\n- [ ] won't fix\n- [ ] _your answer_\n\n## Fixed automatically",
     )
     expect(doc.match(/fix as suggested/g)).toHaveLength(1)
-    expect(unansweredNotes(doc)).toEqual([{ title: "a", line: 5, reason: "tick exactly one box" }])
+    expect(unansweredNotes(doc)).toEqual([{ title: "a", line: 7, reason: "tick exactly one box" }])
   })
 
   it("leaves out the notes already answered", () => {
@@ -1392,7 +1452,7 @@ describe("composedReview", () => {
       noise: [],
       answered: new Set(["a"]),
     })
-    expect(doc).not.toContain("## a\n")
+    expect(doc).not.toContain("# a\n")
     expect(doc).toContain("## a (2)\n\n2")
   })
 })
@@ -1607,3 +1667,17 @@ describe("rankedReview", () => {
     expect(unranked(unranked(review))).toBe(unranked(review))
   })
 })
+
+describe("impact prompts", () => {
+  it.each([
+    ["triage", () => renderText(() => designTriagePrompt("abc123"))],
+    ["design.impact", () => renderText(() => designImpactPrompt())],
+  ])("[AC-2] the %s prompt asks for a `## Impact` section", (_n, prompt) => {
+    expect(prompt()).toContain("`## Impact`")
+  })
+
+  it("[AC-3] the triage prompt names where a return lap reads the human's Impact edit", () => {
+    const p = renderText(() => designTriagePrompt("abc123")).replace(/\s+/g, " ")
+    expect(p).toContain("`git log -1 -p -- .gtd/REQUIREMENTS.md`")
+  })
+})