@windyroad/retrospective 0.21.4 → 0.22.0-preview.479

package/.claude-plugin/plugin.json

@@ -66,5 +66,5 @@
     }
   },
   "name": "wr-retrospective",
-  "version": "0.21.4"
+  "version": "0.22.0"
 }

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@windyroad/retrospective",
-  "version": "0.21.4",
+  "version": "0.22.0-preview.479",
   "description": "Session retrospectives that update briefings and create problem tickets",
   "bin": {
     "windyroad-retrospective": "./bin/install.mjs"

package/skills/run-retro/SKILL.md

@@ -434,7 +434,13 @@ Every codification candidate identified in Step 2 flows through a **two-stage fl
 For every codifiable observation identified in Step 2:
 
 1. **Apply P016 concern-boundary analysis**: if the observation covers multiple independent concerns, split into N observations before ticketing. One ticket per concern.
-2. **Invoke `/wr-itil:manage-problem`** via the Skill tool to create a problem ticket. The observation text becomes the ticket Description; the retro narrative populates the Root Cause Analysis; the `## Related` section cites this retro run. (Once the ADR-032 `capture-*` background sibling ships for manage-problem, Stage 1 can delegate to `/wr-itil:capture-problem` instead so ticketing runs out of the foreground turn; same contract, different invocation mode.)
+2. **Classify the observation per the P342 trust-boundary taxonomy** (mirror of `/wr-itil:work-problems` Step 5 iter-prompt body's classification — same trust-boundary fires whether retro runs in iter context OR standalone in main turn):
+   - **Recurring class-of-behaviour observation** (sibling iters / sessions hit same pattern; SKILL-contract drift; hook misbehaviour; framework-gap; pipeline instability with concrete fix path): **mechanical-auto-ticket** via `/wr-itil:capture-problem` (or `/wr-itil:manage-problem` if capture sibling not yet available). This is the mechanical-stage carve-out per **Step 4a precedent** (verification close-on-evidence — same trust-boundary). Per ADR-013 Rule 5 (policy-authorised silent proceed) + ADR-044 framework-resolution boundary: the retro IS the system designed to mechanically observe recurring patterns; ticketing them is mechanical, not user-decision.
+   - **Direction-setting observation** (genuine user-judgment-bound question — design choice, deviation-approval, framework boundary): route to `outstanding_questions` queue when retro runs inside an AFK iter (via the `ITERATION_SUMMARY.outstanding_questions` schema documented in `packages/itil/skills/work-problems/SKILL.md` Step 5); surface at retro-end interactively when retro runs standalone in main turn. These observations preserve the user's authority surface and MUST NOT auto-ticket.
+   - **Ambiguous** (retro cannot cleanly distinguish recurring-class from direction-setting): **default to mechanical-auto-ticket** per the P342 trust-boundary asymmetry. The ticket lifecycle (`/wr-itil:manage-problem` Step 9d / `/wr-itil:review-problems` Step 4) will surface any embedded direction-setting question through the standard problem-review flow. Defaulting to queue would re-introduce the silent-queue-accumulation hazard P342 closes; defaulting to ticket has zero observation-drop risk.
+
+   This is silent agent judgement — no `AskUserQuestion` per observation. The classification taxonomy is framework-resolved per ADR-044; per-observation `AskUserQuestion` would re-route mechanical decisions back to the user (lazy-deferral surface per Step 2d Ask Hygiene Pass). The work-problems Step 5 iter-prompt body carries the symmetric mirror — both surfaces use the same taxonomy.
+3. **Invoke `/wr-itil:manage-problem`** via the Skill tool to create a problem ticket. The observation text becomes the ticket Description; the retro narrative populates the Root Cause Analysis; the `## Related` section cites this retro run. (Once the ADR-032 `capture-*` background sibling ships for manage-problem, Stage 1 can delegate to `/wr-itil:capture-problem` instead so ticketing runs out of the foreground turn; same contract, different invocation mode.)
 
 **ADR-032 note**: Stage 1 is a legitimate **foreground-spawns-N-background fanout** pattern — run-retro's foreground context spawns one background capture invocation per observation (when the background sibling exists). ADR-032's Confirmation section must carry this case; cite `ADR-032` (`docs/decisions/032-governance-skill-invocation-patterns.proposed.md`) explicitly when the background path lands.
 

package/skills/run-retro/test/run-retro-step-4b-retro-auto-ticket-carveout.bats

@@ -0,0 +1,115 @@
+#!/usr/bin/env bats
+#
+# packages/retrospective/skills/run-retro/test/run-retro-step-4b-retro-auto-ticket-carveout.bats
+#
+# P342: run-retro Step 4b stage classification must mirror the trust-
+# boundary established at Step 4a (verification close-on-evidence) — so
+# the same mechanical-stage carve-out applies whether retro fires in iter
+# context (work-problems Step 5 iter-prompt) OR standalone in main turn.
+#
+# The fix shape amends Step 4b stage classification:
+#   - Mechanical-auto-ticket path: recurring class-of-behaviour /
+#     SKILL-contract drift / hook misbehaviour / framework-gap → Stage 1
+#     auto-creates a ticket via /wr-itil:capture-problem (or
+#     manage-problem if capture sibling not yet available).
+#   - Direction-setting-queue path: genuine user-judgment-bound questions
+#     (design choice, deviation-approval, framework boundary) → queued
+#     as outstanding_questions for orchestrator-level Step 2.5 surface
+#     when retro runs inside an AFK iter; surfaced at retro end when
+#     standalone.
+#   - Ambiguous → default to mechanical-auto-ticket (per P342 trust-
+#     boundary asymmetry preventing silent queue accumulation).
+#
+# Contract-assertion tests per ADR-037 Permitted Exception (structural
+# checks on prose contract; behavioural harness for SKILL.md pending
+# P081 Phase 2 / P012).
+#
+# @problem P342
+# @adr ADR-044 (Decision-Delegation Contract — mechanical-stage carve-out per Step 4a precedent)
+# @adr ADR-013 Rule 5 (policy-authorised silent proceed for capture-* on retro path)
+# @adr ADR-014 (governance skills commit own work — capture-problem commits per ticket)
+# @adr ADR-032 (foreground-spawns-N-background fanout pattern already documented for Stage 1)
+# @jtbd JTBD-006 (durable WSJF-ranked backlog accumulation)
+# @jtbd JTBD-201 (audit-trail integrity)
+
+SKILL_FILE="${BATS_TEST_DIRNAME}/../SKILL.md"
+
+setup() {
+  [ -f "$SKILL_FILE" ]
+}
+
+# ── Step 4b carve-out subsection presence ───────────────────────────────────
+
+@test "Step 4b P342: SKILL.md names the mechanical-auto-ticket vs direction-setting-queue split" {
+  # The fix establishes a classification split at Stage 1 routing.
+  # Mechanical-class observations route to auto-ticket; direction-class
+  # to outstanding_questions queue.
+  run grep -F "mechanical-auto-ticket" "$SKILL_FILE"
+  [ "$status" -eq 0 ]
+}
+
+@test "Step 4b P342: SKILL.md cites Step 4a precedent for the mechanical-stage carve-out" {
+  # The carve-out's authority is the Step 4a verification close-on-
+  # evidence precedent. Cite explicitly so the trust-boundary symmetry
+  # is discoverable.
+  run grep -nE 'Step 4a precedent|same trust[- ]boundary.*Step 4a|Step 4a.*precedent' "$SKILL_FILE"
+  [ "$status" -eq 0 ]
+}
+
+@test "Step 4b P342: SKILL.md names recurring class-of-behaviour as auto-ticket route" {
+  # Recurring class-of-behaviour / SKILL-contract drift / hook
+  # misbehaviour / framework-gap → auto-ticket. The taxonomy must be
+  # documented at Step 4b so the same classification fires whether retro
+  # runs in iter context OR standalone.
+  run grep -nE 'recurring class-of-behaviour|class-of-behaviour observation' "$SKILL_FILE"
+  [ "$status" -eq 0 ]
+}
+
+@test "Step 4b P342: SKILL.md names direction-setting as outstanding_questions route" {
+  # Direction-setting (design choice, deviation-approval, framework
+  # boundary) → outstanding_questions when retro runs inside an AFK
+  # iter; surfaced at retro end when standalone.
+  run grep -nE 'direction-setting.*outstanding_questions|Direction-setting observation' "$SKILL_FILE"
+  [ "$status" -eq 0 ]
+}
+
+@test "Step 4b P342: SKILL.md names ambiguous default-to-auto-ticket asymmetry" {
+  # Per P342 trust-boundary asymmetry: ambiguous → default to auto-
+  # ticket. This prevents the silent-queue accumulation P342 was filed
+  # to close.
+  run grep -nE 'Ambiguous.*auto-ticket|default to auto-ticket|ambiguous.*default.*ticket' "$SKILL_FILE"
+  [ "$status" -eq 0 ]
+}
+
+# ── Cross-reference to work-problems Step 5 iter-prompt carve-out ──────────
+
+@test "Step 4b P342: SKILL.md cross-references work-problems Step 5 iter-prompt symmetry" {
+  # The work-problems iter-prompt amendment is the sibling locus — the
+  # same trust-boundary fires in both surfaces. Cross-reference so the
+  # symmetry is discoverable from either side.
+  run grep -nE 'work-problems Step 5|work-problems.*iter-prompt|Step 5 iter-prompt' "$SKILL_FILE"
+  [ "$status" -eq 0 ]
+}
+
+# ── ADR-044 framework-resolution boundary citation ─────────────────────────
+
+@test "Step 4b P342: SKILL.md cites ADR-044 as the framework-resolution authority for the carve-out" {
+  run grep -F "ADR-044" "$SKILL_FILE"
+  [ "$status" -eq 0 ]
+}
+
+# ── ADR-014 commit ownership preserved ─────────────────────────────────────
+
+@test "Step 4b P342: SKILL.md preserves ADR-014 capture-* commit ownership for auto-ticket path" {
+  # capture-problem (or manage-problem fallback) commits per ADR-014;
+  # run-retro does not commit auto-ticket creates.
+  run grep -F "ADR-014" "$SKILL_FILE"
+  [ "$status" -eq 0 ]
+}
+
+# ── P342 ticket cross-reference ─────────────────────────────────────────────
+
+@test "Step 4b P342: SKILL.md cites P342 as the trust-boundary mirror ticket" {
+  run grep -nE 'P342\b' "$SKILL_FILE"
+  [ "$status" -eq 0 ]
+}