@pmelab/gtd 52.0.0 → 53.0.0

package/src/workflows/feature/refactor-absorbs.feature

@@ -3,10 +3,10 @@
 Feature: a refactor that carried the whole feature still gets a quality lap over all of it
 
   With GTD_FLAG_RISK_TIER off, a refactor-first
-  refactor can land everything the requirements ask for. The feature slice's
+  refactor can land everything the requirements ask for. The feature's
   implement turn then has only the naive pass's acceptance tests to carry
-  over, and with no code but tests changed since the refactor the feature
-  slice's quality lap covers the whole change, refactor included.
+  over, and with no code but tests changed since the refactor the feature's
+  quality lap covers the whole change, refactor included.
 
   Background:
     Given the shipped workflows, as gtd install scaffolds them
@@ -37,6 +37,10 @@ Feature: a refactor that carried the whole feature still gets a quality lap over
 
       - `subtract(a, b)` returns `a - b`.
 
+      ## Impact
+
+      - the greeting module
+
       ## Acceptance Criteria
 
       - AC-1: the change behaves as described.
@@ -128,8 +132,3 @@ Feature: a refactor that carried the whole feature still gets a quality lap over
       gtd(check): implement.closing → build.quality.correctness.reviewing
       gtd(agent): build.quality.correctness.reviewing → build.review.reviewing
       """
-    And the output lines of gtd "slices" start with:
-      """
-      refactor 
-      feature 
-      """

package/src/workflows/feature/refactor-first.feature

@@ -36,6 +36,10 @@ Feature: a refactor-first judgment walks one fixed path to the review gate
 
       - `subtract(a, b)` returns `a - b`.
 
+      ## Impact
+
+      - the greeting module
+
       ## Acceptance Criteria
 
       - AC-1: the change behaves as described.
@@ -156,9 +160,4 @@ Feature: a refactor-first judgment walks one fixed path to the review gate
     And the commit "gtd(agent): seam.naive → seam.check" has a body containing "Gtd-Kept:"
     And the commit "gtd(judge): seam.judge → seam.note" has a body containing "\"answer\":\"refactor-first\""
     And no commit body contains "Gtd-Approved:"
-    And the output lines of gtd "slices" start with:
-      """
-      refactor 
-      feature 
-      """
     And no commit body contains "Gtd-Workflow:"

package/src/workflows/feature/refactor-lens-models.feature

@@ -29,6 +29,10 @@ Feature: the refactor review's lenses take their model hints from refactorQualit
 
       - `subtract(a, b)` returns `a - b`.
 
+      ## Impact
+
+      - the greeting module
+
       ## Acceptance Criteria
 
       - AC-1: the change behaves as described.

package/src/workflows/feature/refactor-rebuild.feature

@@ -37,6 +37,10 @@ Feature: refactor-first writes a refactor note, refactors on its own, then rebui
 
       - `subtract(a, b)` returns `a - b`.
 
+      ## Impact
+
+      - the greeting module
+
       ## Acceptance Criteria
 
       - AC-1: subtract returns the difference.
@@ -193,7 +197,6 @@ Feature: refactor-first writes a refactor note, refactors on its own, then rebui
     Then the process rests at "implement.building"
     And "src/calc.ts" contains "op(a, b, 1)"
     And "src/calc.ts" does not contain "subtract"
-    And the output of gtd "slices" contains "refactor "
 
   Scenario: the rebuild lands only once it carries the naive pass's acceptance tests, green
     Given step "seam.note" writes ".gtd/REFACTOR-NOTE.md" with:
@@ -281,8 +284,3 @@ Feature: refactor-first writes a refactor note, refactors on its own, then rebui
       gtd(check): implement.health.check → implement.closing
       gtd(check): implement.closing → build.review.reviewing
       """
-    And the output lines of gtd "slices" start with:
-      """
-      refactor
-      feature
-      """

package/src/workflows/feature/refactor-review.feature

@@ -1,9 +1,9 @@
 @inmem
-Feature: the refactor slice is reviewed on its own before the feature is implemented
+Feature: the refactor is reviewed on its own before the feature is implemented
 
   After a refactor-first judgment the refactor is built, quality-lapped and
-  reviewed in its own slice. Review edits re-run the refactor, never the design;
-  the feature slice that follows is quality-lapped from the refactor's
+  reviewed on its own. Review edits re-run the refactor, never the design;
+  the feature that follows is quality-lapped from the refactor's
   sign-off, then reviewed. The flags are pinned so the per-flag test runs never move it off
   its path.
 
@@ -48,6 +48,10 @@ Feature: the refactor slice is reviewed on its own before the feature is impleme
 
       - `subtract(a, b)` returns `a - b`.
 
+      ## Impact
+
+      - the greeting module
+
       ## Acceptance Criteria
 
       - AC-1: the change behaves as described.
@@ -146,6 +150,10 @@ Feature: the refactor slice is reviewed on its own before the feature is impleme
 
       - `subtract(a, b)` returns `a - b`.
 
+      ## Impact
+
+      - the greeting module
+
       ## Acceptance Criteria
 
       - AC-1: the change behaves as described.
@@ -245,6 +253,10 @@ Feature: the refactor slice is reviewed on its own before the feature is impleme
 
         - `subtract(a, b)` returns `a - b`.
 
+        ## Impact
+
+        - the greeting module
+
         ## Acceptance Criteria
 
         - AC-1: the change behaves as described.
@@ -602,7 +614,7 @@ Feature: the refactor slice is reviewed on its own before the feature is impleme
       And ".gtd/REFACTOR.md" does not exist
       And ".gtd/REVIEW.md" contains "A: add and subtract differ only by sign."
 
-    Scenario: a no-change landing after the in-gate reply signs off and the feature slice follows
+    Scenario: a no-change landing after the in-gate reply signs off and the feature follows
       Given step "implement.building" writes "src/calc.ts" with:
         """
         export const op = (a: number, b: number, sign: number) => a + sign * b
@@ -685,7 +697,7 @@ Feature: the refactor slice is reviewed on its own before the feature is impleme
       Then the process rests at "build.review.reviewing"
       And "src/calc.ts" contains "subtract"
 
-    Scenario: a praise-only one-shot footnote signs off and the feature slice follows
+    Scenario: a praise-only one-shot footnote signs off and the feature follows
       Given step "implement.building" writes "src/calc.ts" with:
         """
         export const op = (a: number, b: number, sign: number) => a + sign * b
@@ -967,7 +979,7 @@ Feature: the refactor slice is reviewed on its own before the feature is impleme
       Then driving the process is refused with "refactor-feedback"
       And driving the process is refused with "hand-edited src/calc.ts"
 
-    Scenario: a praise note and a concluded thread on a settled round sign off and the feature slice follows
+    Scenario: a praise note and a concluded thread on a settled round sign off and the feature follows
       Given step "implement.building" writes "src/calc.ts" with:
         """
         export const op = (a: number, b: number, sign: number) => a + sign * b
@@ -1054,7 +1066,7 @@ Feature: the refactor slice is reviewed on its own before the feature is impleme
       Then the process rests at "build.review.reviewing"
       And "src/calc.ts" contains "subtract"
 
-    Scenario: a praise-only refactor review still signs off and the feature slice follows
+    Scenario: a praise-only refactor review still signs off and the feature follows
       Given step "implement.building" writes "src/calc.ts" with:
         """
         export const op = (a: number, b: number, sign: number) => a + sign * b
@@ -1087,7 +1099,7 @@ Feature: the refactor slice is reviewed on its own before the feature is impleme
       Then the process rests at "build.review.reviewing"
       And "src/calc.ts" contains "subtract"
 
-    Scenario: a nit fix in the refactor slice that changes a non-stray is refused, though a glob stray widens the write access
+    Scenario: a nit fix in the refactor that changes a non-stray is refused, though a glob stray widens the write access
       Given step "seam.naive" writes "src/*.ts" with:
         """
         export const literalGlobName = 2
@@ -1118,7 +1130,7 @@ Feature: the refactor slice is reviewed on its own before the feature is impleme
       When gtd lands
       Then driving the process is refused with "refactor-scope"
 
-    Scenario: the nit-fix turn in the refactor slice may write the strays and .gtd/** only
+    Scenario: the nit-fix turn in the refactor may write the strays and .gtd/** only
       Given step "seam.naive" writes "src/*.ts" with:
         """
         export const literalGlobName = 2
@@ -1173,7 +1185,7 @@ Feature: the refactor slice is reviewed on its own before the feature is impleme
         """
       And no commit body contains "refactor.build.review.composing"
 
-  Rule: with a quality lens on both slices
+  Rule: with a quality lens on both the refactor and the feature
 
     Background:
       Given an environment variable "GTD_REFACTORQUALITYREVIEWS" set to "correctness"
@@ -1210,6 +1222,10 @@ Feature: the refactor slice is reviewed on its own before the feature is impleme
 
         - `subtract(a, b)` returns `a - b`.
 
+        ## Impact
+
+        - the greeting module
+
         ## Acceptance Criteria
 
         - AC-1: the change behaves as described.
@@ -1259,7 +1275,7 @@ Feature: the refactor slice is reviewed on its own before the feature is impleme
         """
       And step "refactor.build.review.closing" runs its script
 
-    Scenario: each slice's quality lap reviews from that slice's base
+    Scenario: the refactor's and the feature's quality laps each review from their own base
       When the process is driven
       Then the process rests at "refactor.build.quality.correctness.reviewing"
       And the resting step's content contains the hash of the commit "gtd(agent): seam.note → refactor.deriving"
@@ -1337,6 +1353,10 @@ Feature: the refactor slice is reviewed on its own before the feature is impleme
 
         - `subtract(a, b)` returns `a - b`.
 
+        ## Impact
+
+        - the greeting module
+
         ## Acceptance Criteria
 
         - AC-1: the change behaves as described.

package/src/workflows/feature/refactor.ts

@@ -15,9 +15,8 @@ import {
 } from "../../flows/index.js"
 import { lensEntriesOf } from "../../workflow-kit/index.js"
 import { markFenced } from "../shared/fences.js"
-import { continuing, phase, type Phase } from "../shared/continuation.js"
 import { healthy, type Guard } from "../shared/health.js"
-import { coder, FEEDBACK, HANDOFF, REFACTOR } from "../shared/steps.js"
+import { coder, FEEDBACK, REFACTOR } from "../shared/steps.js"
 import * as t from "../shared/text.js"
 import { defaults, testFileGlobs } from "../shared/vars.js"
 import { buildTail, reUnwind, type FeedbackTarget } from "../shared/review.js"
@@ -264,12 +263,7 @@ const guardOf =
   }
 
 /** Derive, build, check and close one refactor lap; `feedback` is the collected review text of the lap before. */
-const refactorLap = async (
-  p: Plan,
-  guard: Guard,
-  refactor: Phase,
-  feedback?: string,
-): Promise<void> => {
+const refactorLap = async (p: Plan, guard: Guard, feedback?: string): Promise<void> => {
   const { pattern, strays, reference, note } = p
   await run(
     "deriving",
@@ -282,59 +276,32 @@ const refactorLap = async (
     ]),
     { label: "Deriving the refactor package", file: REFACTOR_PACKAGE },
   )
-  const reached = new Set<string>()
-  // Each snapshot is greened before its continuation: every sub-step keeps the old suite green.
-  await continuing(refactor, async (from, fresh) => {
-    await fresh(() =>
-      t.agentWithSkills(
-        "building",
-        t.refactorBuildingPrompt({
-          pkg: REFACTOR_PACKAGE,
-          strays,
-          pattern,
-          reference,
-          feedback: feedback !== undefined,
-          note: note !== undefined,
-          from,
-        }),
-        {
-          label: from === undefined ? "Refactoring" : "Continuing the refactor",
-          model: coder(),
-          system: t.builderSystem(),
-        },
-      ),
-    )
-    // Committed by the snapshot, so a turn stopped again may leave it as it was: present means stopped.
-    const left = read(HANDOFF)
-    guard()
-    const snapshot = head()
-    // The building turns' own changes only: a fix turn reaching a path is not the refactor reaching it.
-    for (const c of changes()) reached.add(c.path)
-    await healthy(
-      async () => {
-        await t.agentWithSkills("fix.suite.fixing", t.refactorFixPrompt(strays, pattern), {
-          label: "Fixing the check",
-          file: FEEDBACK,
-          model: coder(),
-          system: t.builderSystem(),
-        })
-        guard()
-      },
-      { describeGuard: guard, rules: t.refactorRules(strays, pattern) },
-    )
-    if (left === undefined) return { result: undefined }
-    return {
-      result: undefined,
-      handoff: {
-        snapshot,
-        note: left.trim(),
-        open: strays.filter((path) => !reached.has(path)),
-      },
-    }
-  })
-  await run("closing", removeScript([REFACTOR_PACKAGE, HANDOFF]), {
-    label: "Closing out the refactor",
-  })
+  await t.agentWithSkills(
+    "building",
+    t.refactorBuildingPrompt({
+      pkg: REFACTOR_PACKAGE,
+      strays,
+      pattern,
+      reference,
+      feedback: feedback !== undefined,
+      note: note !== undefined,
+    }),
+    { label: "Refactoring", model: coder(), system: t.builderSystem() },
+  )
+  guard()
+  await healthy(
+    async () => {
+      await t.agentWithSkills("fix.suite.fixing", t.refactorFixPrompt(strays, pattern), {
+        label: "Fixing the check",
+        file: FEEDBACK,
+        model: coder(),
+        system: t.builderSystem(),
+      })
+      guard()
+    },
+    { describeGuard: guard, rules: t.refactorRules(strays, pattern) },
+  )
+  await run("closing", removeScript([REFACTOR_PACKAGE]), { label: "Closing out the refactor" })
 }
 
 const strayAccess = (strays: readonly string[]) => ({
@@ -348,7 +315,7 @@ export const refactorFirst = async (
   const p = plan(shape, options)
   if (p === undefined) return false
   return scope({ name: "refactor", access: strayAccess(p.strays) }, async () => {
-    await refactorLap(p, guardOf(p, head()), phase("refactor"))
+    await refactorLap(p, guardOf(p, head()))
     return true
   })
 }
@@ -375,10 +342,9 @@ export const reviewedRefactor = async (
         "gtd land: refactor-feedback: the review feedback changed files but wrote no .gtd/REFACTOR.md — a refactor cannot act on it; revert the change, or `gtd abandon` and restart",
       dropped: t.refactorDroppedRefusal,
     }
-    const refactor = phase("refactor")
     let collected: string | undefined
     for (;;) {
-      await refactorLap(p, guard, refactor, collected)
+      await refactorLap(p, guard, collected)
       // A person's edit at an escalation rest is never refused, as in the lap.
       const outcome = await buildTail(entry, {
         green: true,

package/src/workflows/feature/review-answers.feature

@@ -36,6 +36,10 @@ Feature: every note in the review is answered with one tick
 
       - `subtract(a, b)` returns `a - b`.
 
+      ## Impact
+
+      - the greeting module
+
       ## Acceptance Criteria
 
       - AC-1: subtract(a, b) returns a - b.

package/src/workflows/feature/review-gate.feature

@@ -1,7 +1,7 @@
 @inmem
 Feature: a feature process ends at the human review
 
-  After the last slice's quality lap the process rests at the review gate,
+  After the quality lap the process rests at the review gate,
   `.gtd/REVIEW.md`. A sign-off ends the process. A note asking for a change in
   behaviour starts a feedback lap — design, then implement with no new naive
   pass or seam judge, the hands-on test, the quality lap — and the review
@@ -36,6 +36,10 @@ Feature: a feature process ends at the human review
 
       - `subtract(a, b)` returns `a - b`.
 
+      ## Impact
+
+      - the greeting module
+
       ## Acceptance Criteria
 
       - AC-1: subtract(a, b) returns a - b.
@@ -192,6 +196,10 @@ Feature: a feature process ends at the human review
 
         - `multiply(a, b)` returns `a * b`.
 
+        ## Impact
+
+        - the greeting module
+
         ## Acceptance Criteria
 
         - AC-1: subtract(a, b) returns a - b.
@@ -238,6 +246,10 @@ Feature: a feature process ends at the human review
 
         - `multiply(a, b)` returns `a * b`.
 
+        ## Impact
+
+        - the greeting module
+
         ## Acceptance Criteria
 
         - AC-0: multiply(a, b) returns a * b.

package/src/workflows/feature/reviewer-walkthrough.feature

@@ -31,6 +31,10 @@ Feature: a reviewer walks the human through the feature's code
       """
       # Requirements
 
+      ## Impact
+
+      - the greeting module
+
       ## Acceptance Criteria
 
       - AC-1: subtract(a, b) returns a - b.

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

@@ -50,6 +50,10 @@ Feature: the risk tier is proposed at triage and recorded behind GTD_FLAG_RISK_T
 
         low — trivial copy change
 
+        ## Impact
+
+        - the greeting module
+
         ## Acceptance Criteria
 
         - AC-1: the change behaves as described.
@@ -69,6 +73,10 @@ Feature: the risk tier is proposed at triage and recorded behind GTD_FLAG_RISK_T
 
         high — touches payments
 
+        ## Impact
+
+        - the greeting module
+
         ## Acceptance Criteria
 
         - AC-1: the change behaves as described.
@@ -89,6 +97,10 @@ Feature: the risk tier is proposed at triage and recorded behind GTD_FLAG_RISK_T
 
         ### Subtract
 
+        ## Impact
+
+        - the greeting module
+
         ## Acceptance Criteria
 
         - AC-1: the change behaves as described.
@@ -118,6 +130,10 @@ Feature: the risk tier is proposed at triage and recorded behind GTD_FLAG_RISK_T
         - error rate on /subtract stays flat
         - p95 latency
 
+        ## Impact
+
+        - the greeting module
+
         ## Acceptance Criteria
 
         - AC-1: the change behaves as described.
@@ -142,6 +158,10 @@ Feature: the risk tier is proposed at triage and recorded behind GTD_FLAG_RISK_T
 
         low — trivial copy change
 
+        ## Impact
+
+        - the greeting module
+
         ## Acceptance Criteria
 
         - AC-1: the change behaves as described.
@@ -187,6 +207,10 @@ Feature: the risk tier is proposed at triage and recorded behind GTD_FLAG_RISK_T
 
         ### Subtract
 
+        ## Impact
+
+        - the greeting module
+
         ## Acceptance Criteria
 
         - AC-1: the change behaves as described.
@@ -233,6 +257,10 @@ Feature: the risk tier is proposed at triage and recorded behind GTD_FLAG_RISK_T
 
         - error rate stays flat
 
+        ## Impact
+
+        - the greeting module
+
         ## Acceptance Criteria
 
         - AC-1: the change behaves as described.
@@ -258,6 +286,10 @@ Feature: the risk tier is proposed at triage and recorded behind GTD_FLAG_RISK_T
 
         low — trivial
 
+        ## Impact
+
+        - the greeting module
+
         ## Acceptance Criteria
 
         - AC-1: the change behaves as described.

package/src/workflows/feature/seam-rejection.feature

@@ -35,6 +35,10 @@ Feature: rejected naive passes walk one fixed path through the steer gate to the
 
       - `subtract(a, b)` returns `a - b`.
 
+      ## Impact
+
+      - the greeting module
+
       ## Acceptance Criteria
 
       - AC-1: the change behaves as described.
@@ -76,6 +80,10 @@ Feature: rejected naive passes walk one fixed path through the steer gate to the
 
       - `subtract(a, b)` returns `a - b`, in the calculator module.
 
+      ## Impact
+
+      - the greeting module
+
       ## Acceptance Criteria
 
       - AC-1: the change behaves as described.
@@ -128,8 +136,4 @@ Feature: rejected naive passes walk one fixed path through the steer gate to the
     And the commit "gtd(judge): seam.judge → seam.steer" has a body containing "\"answer\":\"reject\""
     And the commit "gtd(judge): seam.judge → implement.applying" has a body containing "\"answer\":\"land\""
     And no commit body contains "Gtd-Approved:"
-    And the output lines of gtd "slices" start with:
-      """
-      feature 
-      """
     And no commit body contains "Gtd-Workflow:"

package/src/workflows/feature/seam.feature

@@ -45,6 +45,10 @@ Feature: the naive pass and seam judge decide how a feature is built
 
       - `subtract(a, b)` returns `a - b`.
 
+      ## Impact
+
+      - the greeting module
+
       ## Acceptance Criteria
 
       - AC-1: the change behaves as described.
@@ -234,6 +238,10 @@ Feature: the naive pass and seam judge decide how a feature is built
 
       - `subtract(a, b)` returns `a - b`, in a new module.
 
+      ## Impact
+
+      - the greeting module
+
       ## Acceptance Criteria
 
       - AC-1: the change behaves as described.