@windyroad/itil 0.61.1 → 0.61.2

package/skills/update-upstream/test/update-upstream-contract.bats

@@ -1,375 +0,0 @@
-#!/usr/bin/env bats
-
-# tdd-review: structural-permitted (justification: P012 skill testing
-#   harness scope is open; no behavioural alternative for SKILL.md prose-
-#   template structural assertions today. ADR-052 Migration clause permits
-#   structural retrofit-via-justification when the linked harness-gap ticket
-#   has not yet shipped the primitives. The paired promptfoo eval at
-#   packages/itil/skills/update-upstream/eval/promptfooconfig.yaml carries
-#   the behavioural Tier-A/B coverage per ADR-075 Amendment 2026-06-02.
-#   Provenance: P080 (original bidirectional contract) + P363 / ADR-024
-#   amendment 2026-06-22 (inbound-origin verdict dispatch leg) + P363 rework
-#   2026-06-23 (four user directives — LLM-generated verdict, workaround
-#   provenance-credit, visibility-gated anti-leakage, cog-a11y gate),
-#   behaviourally paired by the inbound eval cases in promptfooconfig.yaml.
-#   The lockstep transition-problem/manage-problem grep check is an inherent
-#   copy-not-move drift detector (ADR-010/P362) with no behavioural form.)
-#
-# Doc-lint structural test (Permitted Exception per ADR-005 — structural
-# SKILL.md content checks, not behavioural). Mirrors the doc-lint pattern
-# established in ADR-011 and ADR-027 Confirmation tests and used by the
-# sibling report-upstream contract test.
-#
-# Asserts the update-upstream skill's SKILL.md encodes the contract
-# documented in ADR-024 amendment (P080):
-# - sibling-skill split per ADR-010 amended
-# - reads ## Reported Upstream from the local ticket
-# - three transition templates (Open→KE, KE→Verifying, Verifying→Closed)
-# - composes through external-comms + voice-tone gates (AND, not OR)
-# - within appetite → gh issue comment; on Verifying→Closed also gh issue close
-# - above appetite → silent risk-reduce + re-score; if still above, queue
-#   to ## Queued Upstream Update + outstanding_questions (NOT halt the loop)
-# - back-writes to ## Upstream Lifecycle Updates (append-only log)
-# - ADR-014 single-commit grain when fired from transition-problem Step 7
-
-setup() {
-  REPO_ROOT="$(cd "$(dirname "$BATS_TEST_FILENAME")/../../../../.." && pwd)"
-  SKILL_MD="$REPO_ROOT/packages/itil/skills/update-upstream/SKILL.md"
-  EVAL_CONFIG="$REPO_ROOT/packages/itil/skills/update-upstream/eval/promptfooconfig.yaml"
-  TRANSITION_MD="$REPO_ROOT/packages/itil/skills/transition-problem/SKILL.md"
-  MANAGE_MD="$REPO_ROOT/packages/itil/skills/manage-problem/SKILL.md"
-}
-
-@test "update-upstream: SKILL.md exists" {
-  [ -f "$SKILL_MD" ]
-}
-
-@test "update-upstream: SKILL.md declares sibling-skill split per ADR-010 amended (P080 user direction (a))" {
-  # User direction (a) — new sibling skill, not a mode flag on report-upstream.
-  run grep -F 'sibling' "$SKILL_MD"
-  [ "$status" -eq 0 ]
-  run grep -F 'ADR-010' "$SKILL_MD"
-  [ "$status" -eq 0 ]
-}
-
-@test "update-upstream: SKILL.md cross-references ADR-024 amendment (P080)" {
-  run grep -F 'ADR-024' "$SKILL_MD"
-  [ "$status" -eq 0 ]
-  run grep -iE 'P080' "$SKILL_MD"
-  [ "$status" -eq 0 ]
-}
-
-@test "update-upstream: SKILL.md reads ## Reported Upstream from the local ticket (P080 contract)" {
-  # The skill reads the section that /wr-itil:report-upstream Step 7
-  # writes (per ADR-024 Confirmation criterion 3a). Without this read, the
-  # skill has no upstream URL to comment on.
-  run grep -F '## Reported Upstream' "$SKILL_MD"
-  [ "$status" -eq 0 ]
-}
-
-@test "update-upstream: SKILL.md documents three transition templates (Open→KE, KE→Verifying, Verifying→Closed)" {
-  # Each transition has its own template — investigation findings vs release
-  # info vs closure thanks. ADR-024 amendment authority pins all three.
-  run grep -iE 'Open[[:space:]]*[—\-][[:space:]]*[>›]?[[:space:]]*Known Error|Open[[:space:]]*→[[:space:]]*Known Error' "$SKILL_MD"
-  [ "$status" -eq 0 ]
-  run grep -iE 'Known Error[[:space:]]*[—\-][[:space:]]*[>›]?[[:space:]]*Verification Pending|Known Error[[:space:]]*→[[:space:]]*Verification Pending' "$SKILL_MD"
-  [ "$status" -eq 0 ]
-  run grep -iE 'Verification Pending[[:space:]]*[—\-][[:space:]]*[>›]?[[:space:]]*Closed|Verification Pending[[:space:]]*→[[:space:]]*Closed' "$SKILL_MD"
-  [ "$status" -eq 0 ]
-}
-
-@test "update-upstream: SKILL.md composes through external-comms gate (P080 user direction (b) + ADR-028)" {
-  # User direction (b) — risk-gated. The gate is wr-risk-scorer:external-comms
-  # per ADR-028's third-evaluator extension point (the same gate the post-
-  # P270 amendment uses for the initial-filing path).
-  run grep -F 'wr-risk-scorer:external-comms' "$SKILL_MD"
-  [ "$status" -eq 0 ]
-}
-
-@test "update-upstream: SKILL.md composes through voice-tone gate (P080 user direction (b) + ADR-028)" {
-  # User direction (b) — voice-tone gated. The gate is wr-voice-tone:agent
-  # firing on gh issue comment / gh issue close per ADR-028.
-  run grep -F 'ADR-028' "$SKILL_MD"
-  [ "$status" -eq 0 ]
-  run grep -iE 'voice-tone gate' "$SKILL_MD"
-  [ "$status" -eq 0 ]
-}
-
-@test "update-upstream: SKILL.md gates compose AND (both must pass) not OR (P080 user direction (c))" {
-  # User direction (c) — same external-comms gate composition pattern as
-  # inbound (P079). That pattern is AND, not OR — one fail blocks the post.
-  # The SKILL.md must frame composition explicitly so adopters do not read
-  # it as independent gates.
-  run grep -iE 'voice-tone fail.*queue|voice-tone.*FAIL.*queue|FAIL.*identically' "$SKILL_MD"
-  [ "$status" -eq 0 ]
-}
-
-@test "update-upstream: SKILL.md within-appetite posts via gh issue comment" {
-  # Within-appetite path posts via gh issue comment with the matched URL's
-  # issue number — same gh CLI surface as report-upstream Step 5c.
-  run grep -F 'gh issue comment' "$SKILL_MD"
-  [ "$status" -eq 0 ]
-}
-
-@test "update-upstream: SKILL.md Verifying→Closed also runs gh issue close (symmetric audit trail)" {
-  # JTBD-201 symmetric audit trail outcome — when the local ticket goes
-  # .closed.md the upstream issue must also close. Without gh issue close
-  # the upstream tracker accumulates stale-looking open issues.
-  run grep -F 'gh issue close' "$SKILL_MD"
-  [ "$status" -eq 0 ]
-}
-
-@test "update-upstream: SKILL.md above-appetite saves to ## Queued Upstream Update (not the report rename)" {
-  # ADR-024 amendment (P080) — the lifecycle-update queue section is named
-  # ## Queued Upstream Update, distinct from the initial-filing ## Queued
-  # Upstream Report (renamed per the 2026-06-04 second-amendment leaf (c)).
-  # Distinct sections so a single local ticket can carry both without collision.
-  run grep -F '## Queued Upstream Update' "$SKILL_MD"
-  [ "$status" -eq 0 ]
-}
-
-@test "update-upstream: SKILL.md above-appetite queues outstanding_questions (does NOT halt the orchestrator per P352)" {
-  # P352 queue-and-continue — same shape as the post-P270 initial-filing
-  # path. The above-appetite branch must NOT halt the loop; the queued
-  # entry surfaces at the existing batched-AskUserQuestion end-of-loop gate.
-  run grep -iE 'outstanding_questions|queue-and-continue|P352' "$SKILL_MD"
-  [ "$status" -eq 0 ]
-}
-
-@test "update-upstream: SKILL.md silent risk-reduce + re-score is mechanical per ADR-044 (no per-iter AskUserQuestion)" {
-  # ADR-044 framework-resolution boundary — the risk-reduce + re-score is
-  # silent (mechanical); per-iter AskUserQuestion for risk-reduce vocabulary
-  # is the lazy-deferral anti-pattern P132 closes.
-  run grep -iE 'silent risk-reduce|mechanical.*ADR-044|ADR-044.*mechanical' "$SKILL_MD"
-  [ "$status" -eq 0 ]
-}
-
-@test "update-upstream: SKILL.md back-writes to ## Upstream Lifecycle Updates (append-only log)" {
-  # The audit-trail back-write mirrors the ## Reported Upstream back-write
-  # in report-upstream Step 7 (per ADR-024 Confirmation criterion 3a),
-  # extended to log every transition.
-  run grep -F '## Upstream Lifecycle Updates' "$SKILL_MD"
-  [ "$status" -eq 0 ]
-  run grep -iE 'append-only|never overwritten|append.*entry' "$SKILL_MD"
-  [ "$status" -eq 0 ]
-}
-
-@test "update-upstream: SKILL.md AFK behaviour summary table exists (architect review)" {
-  # Mirrors report-upstream's AFK behaviour summary table — the canonical
-  # place where adopters look up the AFK routing for each branch.
-  run grep -F 'AFK behaviour summary' "$SKILL_MD"
-  [ "$status" -eq 0 ]
-}
-
-@test "update-upstream: SKILL.md cites ADR-014 single-commit grain (transition + back-write + upstream post)" {
-  # When fired from transition-problem Step 7, the upstream comment +
-  # back-write + ticket rename + README refresh ride the SAME commit per
-  # ADR-014 single-commit grain.
-  run grep -F 'ADR-014' "$SKILL_MD"
-  [ "$status" -eq 0 ]
-  run grep -iE 'single-commit|same single commit|same commit' "$SKILL_MD"
-  [ "$status" -eq 0 ]
-}
-
-@test "update-upstream: SKILL.md no-op exit on missing ## Reported Upstream (cheap unconditional firing)" {
-  # The trigger fires on every transition unconditionally; the no-op exit
-  # absorbs the misses (most tickets have no ## Reported Upstream). Without
-  # the no-op exit, every transition pays a decision cost.
-  run grep -iE 'no-op exit|nothing to update|exit cleanly' "$SKILL_MD"
-  [ "$status" -eq 0 ]
-}
-
-@test "update-upstream: SKILL.md no-invention rule on template-filling (reporter-trust guard)" {
-  # If a cited local-ticket section is absent, write the explicit "absent"
-  # phrasing rather than invent content. The risk + voice-tone gates cannot
-  # guard against invented technical claims; the no-invention rule does.
-  run grep -iE 'no.invention|do not invent|never invent|rather than invent|absent.*phrasing|"absent"' "$SKILL_MD"
-  [ "$status" -eq 0 ]
-}
-
-@test "update-upstream: SKILL.md annotates JTBD-301 + JTBD-001 (jtbd-lead review)" {
-  # JTBD-301 plugin-user reporter feedback loop AND JTBD-001 solo-developer
-  # without-slowing-down are the load-bearing JTBDs. The jtbd-lead review
-  # required both annotations.
-  run grep -F 'JTBD-301' "$SKILL_MD"
-  [ "$status" -eq 0 ]
-  run grep -F 'JTBD-001' "$SKILL_MD"
-  [ "$status" -eq 0 ]
-}
-
-@test "update-upstream: SKILL.md documents transition-problem Step 7 trigger wire-in (P080 invocation contract)" {
-  # The advisory subsection in transition-problem Step 7 fires this skill
-  # when ## Reported Upstream is present. The SKILL.md must reference the
-  # trigger surface so adopters understand the invocation contract.
-  run grep -iE 'transition-problem.*Step 7|Step 7.*transition-problem|fired from.*transition' "$SKILL_MD"
-  [ "$status" -eq 0 ]
-}
-
-# ─── Inbound-origin verdict dispatch leg (P363 / ADR-024 amendment 2026-06-22) ─
-
-@test "update-upstream: SKILL.md documents the inbound-origin verdict dispatch leg (P363)" {
-  # The skill reads the **Origin**: inbound-reported (#NN) field (ADR-076)
-  # in addition to ## Reported Upstream, so inbound reporters get a verdict.
-  run grep -iE 'Inbound-origin verdict dispatch' "$SKILL_MD"
-  [ "$status" -eq 0 ]
-  run grep -F 'inbound-reported (#NN)' "$SKILL_MD"
-  [ "$status" -eq 0 ]
-}
-
-@test "update-upstream: SKILL.md no-op exit requires BOTH surfaces absent (section AND inbound Origin)" {
-  # The dual-direction no-op exit fires only when NEITHER the outbound
-  # ## Reported Upstream section NOR the inbound **Origin** field is present.
-  run grep -iE 'no inbound Origin|NEITHER .*Reported Upstream.*Origin|both are absent' "$SKILL_MD"
-  [ "$status" -eq 0 ]
-}
-
-@test "update-upstream: SKILL.md inbound leg resolves the originating issue on OUR OWN repo (not external upstream)" {
-  # Inbound issues are filed AGAINST us — the comment posts on our own repo,
-  # resolved via gh default / gh repo view, NOT an external upstream owner/repo.
-  run grep -iE 'gh repo view|our own repo|own repo' "$SKILL_MD"
-  [ "$status" -eq 0 ]
-}
-
-@test "update-upstream: SKILL.md inbound leg has an idempotency guard (gh issue view --json comments)" {
-  # Before posting, scan existing comments for the verdict marker
-  # (package@version + commit SHA) and skip if already posted.
-  run grep -F 'gh issue view' "$SKILL_MD"
-  [ "$status" -eq 0 ]
-  run grep -iE 'idempoten' "$SKILL_MD"
-  [ "$status" -eq 0 ]
-}
-
-@test "update-upstream: SKILL.md inbound templates carry the P229 anti-leakage rule (reporter-facing, no internal vocab)" {
-  # Reporter-facing comment bodies on our own repo MUST NOT leak Step IDs,
-  # branch names, classification tokens, or P/ADR/JTBD IDs as carriers of
-  # meaning. The released @windyroad/<pkg>@<version> is the only permitted
-  # structured token. Mirrors ADR-062 safe-and-valid acknowledgement shape.
-  run grep -iE 'anti-leakage|reporter-surface prose|reporter-facing' "$SKILL_MD"
-  [ "$status" -eq 0 ]
-  run grep -F 'P229' "$SKILL_MD"
-  [ "$status" -eq 0 ]
-}
-
-@test "update-upstream: SKILL.md inbound leg keeps the SAME dual gate composition (no weaker path)" {
-  # The inbound comment routes through the SAME external-comms + voice-tone
-  # dual gate as the outbound path — no weaker path for inbound.
-  run grep -iE 'SAME .*external-comms .*voice-tone|same.*dual gate' "$SKILL_MD"
-  [ "$status" -eq 0 ]
-}
-
-@test "update-upstream: SKILL.md inbound Verifying→Closed also closes the own-repo issue (addresses #97-unclosed witness)" {
-  # P211 #97 was silent AND unclosed. The inbound leg runs gh issue close
-  # on the own repo on the Verifying→Closed transition.
-  run grep -iE 'posted-inbound-comment-and-closed|gh issue close .*OWN' "$SKILL_MD"
-  [ "$status" -eq 0 ]
-}
-
-@test "update-upstream: SKILL.md both-direction tickets fire both legs independently" {
-  # A ticket reported upstream AND reported against us inbound fires both
-  # dispatches independently — one above-appetite leg queues only itself.
-  run grep -iE 'Both-direction|both.*legs.*independent|fire .*independently' "$SKILL_MD"
-  [ "$status" -eq 0 ]
-}
-
-@test "update-upstream: SKILL.md back-writes inbound entries direction-tagged so P249 poller is uncontaminated" {
-  # Inbound entries log to ## Upstream Lifecycle Updates (NOT ## Reported
-  # Upstream) so /wr-itil:check-upstream-responses (P249) is not contaminated.
-  run grep -F 'P249' "$SKILL_MD"
-  [ "$status" -eq 0 ]
-  run grep -iE 'inbound.*disclosure path|posted-inbound-comment' "$SKILL_MD"
-  [ "$status" -eq 0 ]
-}
-
-# ─── P363 rework 2026-06-23 — four user directives ────────────────────────────
-
-@test "update-upstream: SKILL.md inbound verdict is LLM-generated per-context, not templated (Directive 1)" {
-  # Directive 1 — templates produce cold, patterned prose; the inbound verdict
-  # is GENERATED from ticket context via per-transition generation prompts the
-  # agent reads at runtime, NOT filled from canned template bodies.
-  run grep -iE 'LLM-generated|generated from the local ticket|generation prompt' "$SKILL_MD"
-  [ "$status" -eq 0 ]
-  run grep -F 'no templates' "$SKILL_MD"
-  [ "$status" -eq 0 ]
-}
-
-@test "update-upstream: SKILL.md O→KE prompt shares the workaround + scans issue comments for provenance (Directive 2/4)" {
-  # Directive 2 — a Known Error comment shares the WORKAROUND, not just root
-  # cause. Directive 4 — provenance: read ## Workaround AND scan the originating
-  # issue's body+comments (gh issue view --json comments,body) to attribute it.
-  run grep -F 'share it with correct PROVENANCE' "$SKILL_MD"
-  [ "$status" -eq 0 ]
-  run grep -F 'comments,body' "$SKILL_MD"
-  [ "$status" -eq 0 ]
-  run grep -F '## Workaround' "$SKILL_MD"
-  [ "$status" -eq 0 ]
-}
-
-@test "update-upstream: SKILL.md O→KE prompt names four provenance branches + credit shape + confirm-exact (Directive 4)" {
-  # The prompt must branch on who provided the workaround and credit per branch,
-  # always confirming the exact validated details.
-  run grep -F 'MAINTAINER-AUTHORED' "$SKILL_MD"; [ "$status" -eq 0 ]
-  run grep -F 'REPORTER-PROVIDED' "$SKILL_MD"; [ "$status" -eq 0 ]
-  run grep -F 'COMMENTER-PROVIDED' "$SKILL_MD"; [ "$status" -eq 0 ]
-  run grep -F 'BOTH-SOURCE' "$SKILL_MD"; [ "$status" -eq 0 ]
-  run grep -iE 'credit the reporter|credit them by @handle|credit.*both' "$SKILL_MD"; [ "$status" -eq 0 ]
-  run grep -iE 'CONFIRM THE EXACT|Confirmed the details|confirm the details' "$SKILL_MD"; [ "$status" -eq 0 ]
-}
-
-@test "update-upstream: SKILL.md inbound anti-leakage is visibility-gated — PUBLIC titled+linked, PRIVATE strict ban (Directive 3)" {
-  # Directive 3 — for PUBLIC repos refer to problems/ADRs/RFCs by title AND link
-  # (never bare ID); for PRIVATE/INTERNAL/indeterminate fall back to strict ban.
-  run grep -F 'gh repo view --json visibility' "$SKILL_MD"; [ "$status" -eq 0 ]
-  run grep -iE 'title AND as a link|titled\+linked' "$SKILL_MD"; [ "$status" -eq 0 ]
-  run grep -iE 'PRIVATE / INTERNAL / indeterminate|strict ban' "$SKILL_MD"; [ "$status" -eq 0 ]
-  # classification tokens / step IDs / internal vocab stay banned regardless.
-  run grep -iE 'Always banned' "$SKILL_MD"; [ "$status" -eq 0 ]
-  # grounded in ADR-055's permalink-progressive-enhancement.
-  run grep -F 'ADR-055' "$SKILL_MD"; [ "$status" -eq 0 ]
-}
-
-@test "update-upstream: SKILL.md inbound gate chain rides cog-a11y first (when-available, P338-gated, do-not-block)" {
-  # Directive 1's "run cog-a11y, risk, voice and tone" — cog-a11y rides FIRST,
-  # gated on P338 (plugin not yet shipped); the chain degrades to the dual gate
-  # today and does NOT block on P338.
-  run grep -iE 'cog-a11y → risk → voice-tone|cognitive-accessibility → external-comms' "$SKILL_MD"; [ "$status" -eq 0 ]
-  run grep -F 'P338' "$SKILL_MD"; [ "$status" -eq 0 ]
-  run grep -iE 'do NOT block|when-available' "$SKILL_MD"; [ "$status" -eq 0 ]
-}
-
-@test "transition-problem + manage-problem Step 7 greps match the inbound Origin alternation in LOCKSTEP (P363/P362)" {
-  # Both lockstep pre-checks must match ^## Reported Upstream OR
-  # ^\*\*Origin\*\*: inbound-reported \(# — drift re-opens the P363 gap on
-  # whichever surface lacks the alternation (ADR-010 copy-not-move / P362).
-  run grep -F 'inbound-reported \(#' "$TRANSITION_MD"
-  [ "$status" -eq 0 ]
-  run grep -F 'inbound-reported \(#' "$MANAGE_MD"
-  [ "$status" -eq 0 ]
-}
-
-# ─── Paired promptfoo eval (ADR-075 Amendment 2026-06-02) ────────────────────
-#
-# The paired promptfoo eval at packages/itil/skills/update-upstream/eval/
-# discharges the R009 prose-floor for this SKILL surface atomically per
-# ADR-075 evidence floor.
-
-@test "update-upstream: paired promptfoo eval config exists (ADR-075 Amendment 2026-06-02)" {
-  [ -f "$EVAL_CONFIG" ]
-}
-
-@test "update-upstream: paired eval uses exec:bash provider per ADR-075 §6 (no API key)" {
-  # ADR-075 §6 — subscription auth via claude -p, no ANTHROPIC_API_KEY.
-  # The provider must be exec:bash invoking ./run-skill-eval.sh.
-  run grep -F "exec:bash ./run-skill-eval.sh" "$EVAL_CONFIG"
-  [ "$status" -eq 0 ]
-}
-
-@test "update-upstream: paired eval ships Tier-B llm-rubric grader for negative clauses" {
-  # Negative clauses (must NOT halt; must NOT invent) cannot be Tier-A
-  # graded — a not-regex for "halt" false-fails the correct "does NOT halt".
-  # The grader script + llm-rubric assertions discharge those clauses.
-  GRADER="$REPO_ROOT/packages/itil/skills/update-upstream/eval/grade-llm-rubric.sh"
-  [ -f "$GRADER" ]
-  [ -x "$GRADER" ]
-  run grep -F 'llm-rubric' "$EVAL_CONFIG"
-  [ "$status" -eq 0 ]
-}

package/skills/work-problem/test/work-problem-contract.bats

@@ -1,294 +0,0 @@
-#!/usr/bin/env bats
-# Contract assertions for /wr-itil:work-problem (P071 split slice 3).
-#
-# This skill hosts the "pick the highest-WSJF ticket and work it" user
-# intent previously hidden behind /wr-itil:manage-problem work. Unlike
-# the plural orchestrator (/wr-itil:work-problems — batch AFK loop),
-# this singular skill runs exactly one ticket per invocation and fires
-# the AskUserQuestion selection prompt in interactive mode. Execution
-# is delegated to /wr-itil:manage-problem <NNN> to avoid forking the
-# per-ticket workflow.
-#
-# Structural assertion — Permitted Exception to the source-grep ban
-# (ADR-005 / P011 / ADR-037 contract-assertion pattern).
-#
-# tdd-review: structural-permitted (justification: SKILL.md prose contract
-# assertions; behavioural skill-runtime harness pending P012 + P081 Phase 2;
-# expected to migrate to behavioural form once the harness exists. Touched
-# during P136 Phase 2 ADR-044 alignment audit per the inline plan's
-# bridge-marker rule.)
-#
-# @problem P071 (originating split)
-# @problem P136 (ADR-044 alignment audit master — Phase 2 work-problem singular)
-# @adr ADR-044 (Decision-Delegation Contract — framework-mediated Prioritisation surface)
-# @jtbd JTBD-001 (enforce governance without slowing down — discoverable surface)
-# @jtbd JTBD-101 (extend the suite with clear patterns — one skill per distinct user intent)
-# @jtbd JTBD-201 (audit trail — selection cites the deciding tie-break rung)
-#
-# Cross-reference:
-#   P071: docs/problems/071-argument-based-skill-subcommands-are-not-discoverable.open.md
-#   P136: docs/problems/136-adr-044-alignment-audit-master.open.md
-#   ADR-010 amended (Skill Granularity section) — split naming + forwarder contract
-#   ADR-013 amended Rule 1 — structured user interaction; framework-resolution narrowing per ADR-044
-#   ADR-013 Rule 6 — AFK non-interactive fallback (Step 4 scope-expansion only)
-#   ADR-014 — governance skills commit their own work (delegated target owns commits)
-#   ADR-032 — governance skill invocation patterns (plural orchestrator delegates here)
-#   ADR-037 — contract-assertion bats pattern
-#   ADR-044 — Decision-Delegation Contract; Step 2 selection is framework-mediated
-
-setup() {
-  SKILL_DIR="$(cd "$(dirname "$BATS_TEST_FILENAME")/.." && pwd)"
-  SKILL_FILE="${SKILL_DIR}/SKILL.md"
-}
-
-@test "SKILL.md exists and has frontmatter" {
-  [ -f "$SKILL_FILE" ]
-  run head -1 "$SKILL_FILE"
-  [ "$status" -eq 0 ]
-  [ "$output" = "---" ]
-}
-
-@test "SKILL.md frontmatter name is wr-itil:work-problem (singular, P071 + ADR-010 amended)" {
-  # Split naming convention per ADR-010 amendment: <verb>-<object>.
-  # CRITICAL: singular "work-problem", NOT plural "work-problems" — the
-  # plural name is reserved for the AFK orchestrator that already exists.
-  # Name coexistence is a P071 acknowledged trade-off.
-  run grep -n "^name: wr-itil:work-problem$" "$SKILL_FILE"
-  [ "$status" -eq 0 ]
-}
-
-@test "SKILL.md frontmatter name is NOT wr-itil:work-problems (regression guard for name collision)" {
-  # Guard against the easy slip of renaming this skill to match the
-  # plural orchestrator. If the name drifts to plural, `/wr-itil:`
-  # autocomplete shows two plural-looking skills side by side and users
-  # cannot tell which is interactive vs batch.
-  run grep -n "^name: wr-itil:work-problems$" "$SKILL_FILE"
-  [ "$status" -ne 0 ]
-}
-
-@test "SKILL.md frontmatter description names the pick-and-run intent (P071)" {
-  # Description must name "pick" / "work" / "highest-WSJF" so Claude
-  # Code autocomplete surfaces the user intent rather than a generic
-  # name. Also must mention "singular" or distinguish from the plural
-  # orchestrator so users pick the right one.
-  run grep -inE "^description:.*(pick|highest.wsjf|work.*problem)" "$SKILL_FILE"
-  [ "$status" -eq 0 ]
-  run grep -inE "^description:.*(singular|one ticket|distinct from.*work-problems)" "$SKILL_FILE"
-  [ "$status" -eq 0 ]
-}
-
-@test "SKILL.md frontmatter allowed-tools includes AskUserQuestion (selection prompt)" {
-  # Step 2 fires AskUserQuestion for the selection in interactive mode
-  # (ADR-013 Rule 1). Without the tool the structured-interaction
-  # contract cannot hold.
-  run grep -nE "^allowed-tools:.*AskUserQuestion" "$SKILL_FILE"
-  [ "$status" -eq 0 ]
-}
-
-@test "SKILL.md frontmatter allowed-tools includes Skill (delegation to manage-problem + review-problems)" {
-  # Step 1 delegates the cache refresh to /wr-itil:review-problems when
-  # stale; Step 3 delegates the per-ticket execution to
-  # /wr-itil:manage-problem <NNN>. Both use the Skill tool per ADR-010's
-  # thin-router discipline.
-  run grep -nE "^allowed-tools:.*Skill" "$SKILL_FILE"
-  [ "$status" -eq 0 ]
-}
-
-@test "SKILL.md documents the singular-vs-plural naming distinction (P071 coexistence)" {
-  # The most common user confusion will be work-problem vs
-  # work-problems. The SKILL.md must name the distinction explicitly
-  # so `/wr-itil:` autocomplete readers can tell them apart. Without
-  # this, users will pick the wrong one and either run a single ticket
-  # when they wanted the AFK loop or vice-versa.
-  run grep -inE "work-problems.*plural|plural.*work-problems|AFK orchestrator" "$SKILL_FILE"
-  [ "$status" -eq 0 ]
-}
-
-@test "SKILL.md delegates execution to /wr-itil:manage-problem (anti-fork discipline)" {
-  # Per ADR-010 thin-router: the split skill owns selection, NOT
-  # execution. The per-ticket investigate/transition/fix/release flow
-  # stays on /wr-itil:manage-problem <NNN>. If this skill re-implements
-  # the execution inline, the deprecation window hardens into a
-  # permanent fork.
-  run grep -inE "/wr-itil:manage-problem" "$SKILL_FILE"
-  [ "$status" -eq 0 ]
-  run grep -inE "delegate.*manage-problem|Skill tool.*manage-problem|manage-problem.*Skill tool" "$SKILL_FILE"
-  [ "$status" -eq 0 ]
-}
-
-@test "SKILL.md defers README.md cache refresh to /wr-itil:review-problems (P062 ownership)" {
-  # P062 makes /wr-itil:review-problems the canonical README.md cache
-  # writer. This skill reads the cache but must NOT rewrite it —
-  # re-implementing the refresh here would break the one-canonical-
-  # writer invariant.
-  run grep -inE "/wr-itil:review-problems" "$SKILL_FILE"
-  [ "$status" -eq 0 ]
-  run grep -inE "(delegate|defer|refresh).*review-problems|review-problems.*(refresh|delegate|defer)" "$SKILL_FILE"
-  [ "$status" -eq 0 ]
-}
-
-@test "SKILL.md reads README.md cache via the git-history freshness test (P031)" {
-  # Same freshness check as list-problems and review-problems — mtime
-  # is unreliable in worktrees per P031. Drift here means false
-  # cache-hit (serving a stale ranking) or false cache-miss (spamming
-  # review refreshes).
-  run grep -inE "docs/problems/README\.md" "$SKILL_FILE"
-  [ "$status" -eq 0 ]
-  run grep -inE "git log.*--format.*README\.md|readme_commit" "$SKILL_FILE"
-  [ "$status" -eq 0 ]
-}
-
-@test "SKILL.md fires the AskUserQuestion selection prompt (ADR-013 Rule 1)" {
-  # Step 2 is the structured-interaction path: single top-WSJF ticket
-  # shown as Recommended, or tied tickets shown as peer options. Never
-  # prose "(a)/(b)/(c)".
-  run grep -inE "AskUserQuestion" "$SKILL_FILE"
-  [ "$status" -eq 0 ]
-  run grep -inE "Next problem|Recommended|tied" "$SKILL_FILE"
-  [ "$status" -eq 0 ]
-}
-
-@test "SKILL.md forbids prose selection fallback (P053 + ADR-013 Rule 1 regression guard)" {
-  # Prose "which would you like?" or "(a)/(b)/(c)" is the anti-pattern
-  # ADR-013 Rule 1 and P053 codify against. The SKILL.md must say
-  # "never" so the contract is legible at the spec level.
-  run grep -inE "never.*prose|never.*\(a\)/\(b\)" "$SKILL_FILE"
-  [ "$status" -eq 0 ]
-}
-
-@test "SKILL.md documents the AFK non-interactive branch (ADR-013 Rule 6)" {
-  # When /wr-itil:work-problems invokes this skill inside an AFK
-  # subagent iteration, AskUserQuestion is unavailable — the selection
-  # has already happened at the orchestrator level. The skill must
-  # degrade gracefully per Rule 6.
-  run grep -inE "AFK|non-interactive|ADR-013 Rule 6|Rule 6" "$SKILL_FILE"
-  [ "$status" -eq 0 ]
-}
-
-@test "SKILL.md documents the scope-expansion AskUserQuestion (3-option shape)" {
-  # Effort drift during investigation or architect review must fire the
-  # standard Continue / Re-rank / Pick-different three-option prompt,
-  # same as /wr-itil:manage-problem's Working a Problem section.
-  run grep -inE "Scope change|scope.expansion" "$SKILL_FILE"
-  [ "$status" -eq 0 ]
-  run grep -inE "Continue with expanded scope|Update problem and re-rank|Pick a different problem" "$SKILL_FILE"
-  [ "$status" -eq 0 ]
-}
-
-@test "SKILL.md runs exactly one ticket per invocation (singular contract)" {
-  # The whole point of the singular skill: one ticket, then stop. The
-  # SKILL.md must say "do not loop automatically" or equivalent so the
-  # contract is explicit. Looping is the plural orchestrator's job.
-  run grep -inE "do not loop automatically|one ticket per invocation|singular" "$SKILL_FILE"
-  [ "$status" -eq 0 ]
-}
-
-@test "SKILL.md does not carry a deprecated-arguments frontmatter flag (clean-split skill)" {
-  # Architect advisory: work-problem is a clean-split skill with no
-  # argument-subcommands itself. ADR-010 amendment's
-  # `deprecated-arguments: true` flag is only valid on host skills with
-  # forwarder routes — work-problem is a forwarder TARGET, not a host.
-  run grep -E "^deprecated-arguments:" "$SKILL_FILE"
-  [ "$status" -ne 0 ]
-}
-
-@test "SKILL.md does not use word-argument subcommand branching (P071 regression guard)" {
-  # The whole point of P071: Claude Code autocomplete does not surface
-  # arguments. A clean-split skill must not reintroduce word-arg
-  # subcommand routing.
-  run grep -inE "If arguments start with|If arguments contain" "$SKILL_FILE"
-  [ "$status" -ne 0 ]
-}
-
-@test "SKILL.md cites P071 and ADR-010 amended (traceability per ADR-037)" {
-  # ADR-037 traceability: the skill spec cites the problem it closes
-  # and the ADR that authorises the split.
-  run grep -inE "P071|ADR-010" "$SKILL_FILE"
-  [ "$status" -eq 0 ]
-}
-
-@test "SKILL.md cites ADR-032 + P077 (AFK orchestrator delegation target)" {
-  # /wr-itil:work-problems Step 5 delegates iterations via the Agent
-  # tool (P077). This skill is the execution unit those iterations
-  # invoke. The cross-reference documents the pair relationship so
-  # future drift to /wr-itil:work-problems doesn't orphan this skill.
-  run grep -inE "P077|ADR-032" "$SKILL_FILE"
-  [ "$status" -eq 0 ]
-}
-
-# ----------------------------------------------------------------------
-# P136 Phase 2 — ADR-044 alignment audit (added 2026-04-27)
-#
-# These assertions land the framework-mediated-selection contract for
-# Step 2 and the ADR-044 category-2 cross-reference for Step 4. They
-# replace the prior assertions that mandated AskUserQuestion-driven
-# selection (which was the lazy-deferral surface ADR-044 was written to
-# close).
-# ----------------------------------------------------------------------
-
-@test "SKILL.md Step 2 specifies framework-mediated selection (ADR-044 Prioritisation)" {
-  # Per ADR-044 Framework-Mediated Surface row "Prioritisation — WSJF
-  # formula + documented tie-breaks (Known Error > Open; smaller effort
-  # first; older reported date). Pick + work." Step 2 must apply the
-  # tie-break ladder mechanically rather than defer to AskUserQuestion.
-  run awk '/^### 2\./,/^### 3\./' "$SKILL_FILE"
-  [ "$status" -eq 0 ]
-  [[ "$output" == *"framework-mediated"* ]] || [[ "$output" == *"tie-break ladder"* ]]
-  [[ "$output" == *"Known Error"* ]]
-  [[ "$output" == *"smaller"* ]]
-  [[ "$output" == *"older reported date"* ]]
-  [[ "$output" == *"ADR-044"* ]]
-}
-
-@test "SKILL.md Step 2 reports the chosen ticket + tie-break rung that decided (JTBD-201 audit-trail)" {
-  # Audit-trail outcome: the agent must cite which rung of the ladder
-  # decided the selection so the choice is reproducible from the README
-  # state at the time of the report.
-  run awk '/^### 2\./,/^### 3\./' "$SKILL_FILE"
-  [ "$status" -eq 0 ]
-  [[ "$output" == *"report"* ]]
-  [[ "$output" == *"tie-break"* ]] || [[ "$output" == *"ladder rung"* ]]
-}
-
-@test "SKILL.md Step 2 documents user-override path via direct NNN invocation" {
-  # Selection is framework-mediated, but the user retains agency via
-  # /wr-itil:work-problem <NNN> direct invocation. Without the
-  # documented escape hatch, the user has no way to redirect except via
-  # post-hoc correction (ADR-044 category 6). The literal-form
-  # assertion guards against the substring trap where /wr-itil:work-
-  # problems (plural) accidentally satisfies a /wr-itil:work-problem
-  # substring check.
-  run awk '/^### 2\./,/^### 3\./' "$SKILL_FILE"
-  [ "$status" -eq 0 ]
-  [[ "$output" == *"/wr-itil:work-problem <NNN>"* ]] || [[ "$output" == *"/wr-itil:work-problem <ID>"* ]] || [[ "$output" == *"\`/wr-itil:work-problem 042\`"* ]]
-}
-
-@test "SKILL.md Step 2 does NOT fire AskUserQuestion as the selection mechanism (regression guard)" {
-  # The AskUserQuestion-driven selection prompt was the lazy-deferral
-  # surface ADR-044 closed. If it returns to Step 2 the lazy-count
-  # metric (Step 2d "Ask Hygiene Pass") will spike. AskUserQuestion may
-  # appear elsewhere (Step 4 scope-expansion) but NOT inside Step 2.
-  run awk '/^### 2\./,/^### 3\./' "$SKILL_FILE"
-  [ "$status" -eq 0 ]
-  ! echo "$output" | grep -qE "Selection via .?AskUserQuestion|Use .header: .Next problem"
-}
-
-@test "SKILL.md Step 4 cross-references ADR-044 category-2 (deviation-approval)" {
-  # Effort growth IS the contradicting evidence against the WSJF score
-  # that ranked this ticket at the top — it's a deviation-approval
-  # surface in the ADR-044 taxonomy. The inline cross-reference makes
-  # the framework-resolution boundary visible at the call site.
-  run awk '/^### 4\./,/^### 5\./' "$SKILL_FILE"
-  [ "$status" -eq 0 ]
-  [[ "$output" == *"ADR-044"* ]]
-  [[ "$output" == *"deviation-approval"* ]] || [[ "$output" == *"category 2"* ]] || [[ "$output" == *"category-2"* ]]
-}
-
-@test "bats file carries the tdd-review: structural-permitted marker (P081 + P136 bridge)" {
-  # Per P136 Phase 2 inline plan: bats touched during the audit get
-  # the structural-permitted marker as the bridge until P081 Phase 2's
-  # canonical retrofit. Without the marker, future TDD agent reviews
-  # will flag the file as a P081 violation.
-  run grep -nE "tdd-review:[[:space:]]+structural-permitted" "${BATS_TEST_FILENAME}"
-  [ "$status" -eq 0 ]
-}