@mmerterden/multi-agent-pipeline 16.20.0 → 16.23.0

package/pipeline/scripts/check-read-size.sh

@@ -0,0 +1,86 @@
+#!/usr/bin/env bash
+# check-read-size.sh  -  PreToolUse read gate for the multi-agent pipeline.
+#
+# The third deterministic hook, and the first on the READ side. The other two
+# (pre-commit-check, agent-guard) inspect what the run writes; this one inspects
+# what it pays to read.
+#
+# The bill it exists for: a phase that reads six 900-line files pays for 5,400
+# lines at that phase's own rung, and the part it needed was a handful of
+# symbols. `offload-ref.sh` already took the other half of that bill - the build
+# log, the diff, the test output - by parking the payload and printing a pointer.
+# Nothing looked at file reads, because nothing could: only a hook sees a tool
+# call before it runs.
+#
+# Contract (Claude Code PreToolUse hook):
+#   - Reads the tool-call JSON on stdin.
+#   - Exit 2  -> BLOCK (reason on stderr, shown to the model).
+#   - Exit 0  -> allow.
+#
+# Modes  -  `prefs.global.bulkRead.mode`:
+#   off       (default) nothing is inspected; the hook is a no-op.
+#   observe   decide and LOG, never block. This is the measurement: run the
+#             pipeline normally and metrics.jsonl carries the read-size
+#             distribution, so the saving can be estimated before anything is
+#             built on top of it.
+#   enforce   block a whole-file read over `minLines` and tell the model to
+#             delegate it to bulk-read.sh instead.
+#
+# Why observe is a real mode and not a debug flag: the honest order is measure,
+# then route. A gate switched straight to enforce has no baseline to be
+# compared against, and "we cut tokens" becomes a claim nothing can check.
+#
+# Why the development phase is exempt by default: in Claude Code, `Edit` requires
+# the same file to have been read first. A gate that blocks reads in Phase 3
+# blocks editing. The gate is for the phases that read to UNDERSTAND.
+#
+# Safety design (fail-OPEN): any internal error, missing python3, missing helper,
+# empty payload -> exit 0. It never executes the inspected command, never writes
+# to the inspected file, makes no network call, and prints no file content.
+
+set -u
+
+HERE="$(cd "$(dirname "$0")" 2>/dev/null && pwd || true)"
+HELPER="$HERE/check-read-size.py"
+[ -f "$HELPER" ] || exit 0
+command -v python3 >/dev/null 2>&1 || exit 0
+
+PAYLOAD="$(cat 2>/dev/null || true)"
+[ -z "$PAYLOAD" ] && exit 0
+
+DECISION="$(printf '%s' "$PAYLOAD" | python3 "$HELPER" 2>/dev/null || true)"
+[ -z "$DECISION" ] && exit 0
+
+VERDICT="$(printf '%s' "$DECISION" | cut -f1)"
+FILE="$(printf '%s' "$DECISION" | cut -f2)"
+LINES="$(printf '%s' "$DECISION" | cut -f3)"
+WHY="$(printf '%s' "$DECISION" | cut -f4)"
+
+# Telemetry is best-effort and never gates the decision. `log-metric.sh` already
+# tolerates an unwritable directory, so the only guard needed here is the
+# script's own presence.
+note() {
+  [ -x "$HERE/log-metric.sh" ] || return 0
+  "$HERE/log-metric.sh" "${MULTI_AGENT_TASK_ID:-unknown}" "${MULTI_AGENT_PHASE:-0}" \
+    "read.$1" lines="${LINES:-0}" reason="${WHY:-}" >/dev/null 2>&1 || true
+}
+
+case "$VERDICT" in
+  OBSERVE)
+    note observed
+    exit 0 ;;
+  BLOCK)
+    note blocked
+    echo "BLOCKED by check-read-size: ${FILE} is ${LINES} lines." >&2
+    echo "Reading it whole costs this phase's rung for a file whose useful part is a few symbols." >&2
+    echo "" >&2
+    echo "Delegate it, then work from what comes back:" >&2
+    echo "  bash \$HOME/.claude/scripts/bulk-read.sh --file '${FILE}' --question '<what you need from it>'" >&2
+    echo "" >&2
+    echo "The summary carries LINE NUMBERS, so a follow-up Read with an offset/limit around the" >&2
+    echo "region you actually need is the intended next step and passes this gate." >&2
+    echo "Full text stays on disk under .multi-agent/refs/ if the summary is not enough." >&2
+    exit 2 ;;
+  *)
+    exit 0 ;;
+esac

package/pipeline/scripts/phase-tracker.sh

@@ -64,6 +64,8 @@ usage:
   phase-tracker.sh meta   <phase_id> <key> <value>
   phase-tracker.sh now    <phase_id> "<text>"
   phase-tracker.sh cost   <phase_id>|total
+  phase-tracker.sh tiles
+  phase-tracker.sh report
   phase-tracker.sh render
 USAGE
   exit 64
@@ -498,6 +500,167 @@ release_state_lock() {
 # Never leave a lock behind if a mutating action dies mid-critical-section.
 trap release_state_lock EXIT
 
+# The cost table is keyed by family ("opus", "sonnet", "gpt-5.6"), but the name
+# an agent has at hand is the full id ("claude-opus-5"). Storing the id verbatim
+# priced every phase at "-" while the tracker looked perfectly healthy, so the
+# family is resolved here, once, and an unresolvable name is said out loud
+# instead of quietly costing nothing.
+normalize_model() {
+  local raw="$1" lower
+  lower=$(printf '%s' "$raw" | tr '[:upper:]' '[:lower:]')
+  case "$lower" in
+    *terra*) printf 'gpt-5.6-terra' ;;
+    *gpt-5.6* | *gpt5.6*) printf 'gpt-5.6' ;;
+    *gpt-5.4* | *gpt5.4*) printf 'gpt-5.4' ;;
+    *opus*) printf 'opus' ;;
+    *sonnet*) printf 'sonnet' ;;
+    *haiku*) printf 'haiku' ;;
+    *fable*) printf 'fable' ;;
+    *) printf '%s' "$raw" ;;
+  esac
+}
+
+# Which CLI tree this copy runs from. The card below is identical on all three
+# hosts, but the WIDGET is not: Claude Code has TaskCreate/TaskUpdate, Codex has
+# update_plan, Copilot CLI has no task UI at all. Every "what to call next" hint
+# has to know where it is, or it names a tool the host does not have.
+SELF_PATH="${BASH_SOURCE[0]:-$0}"
+host_kind() {
+  case "${MULTI_AGENT_HOST:-}" in
+    claude | codex | copilot)
+      printf '%s' "$MULTI_AGENT_HOST"
+      return 0
+      ;;
+  esac
+  case "$SELF_PATH" in
+    */.codex/*) printf 'codex' ;;
+    */.copilot/*) printf 'copilot' ;;
+    *) printf 'claude' ;;
+  esac
+}
+
+# The completion line the contract requires, built from what was actually
+# recorded. Phases with no recorded spend say so rather than printing zeros,
+# which is the difference between "cost unknown" and "cost was nothing".
+narration_line() {
+  local pid="$1" state name idx model t_in t_out usd
+  state="$(load_state)"
+  name=$(echo "$state" | jq -r --arg id "$pid" '(.phases[] | select(.id == $id) | .name) // ""')
+  idx=$(echo "$state" | jq -r --arg id "$pid" '[.phases[]?.id] | index($id) // "-"')
+  model=$(echo "$state" | jq -r --arg id "$pid" '(.phases[] | select(.id == $id) | .model) // ""')
+  t_in=$(echo "$state" | jq -r --arg id "$pid" '(.phases[] | select(.id == $id) | .tokens_in) // 0')
+  t_out=$(echo "$state" | jq -r --arg id "$pid" '(.phases[] | select(.id == $id) | .tokens_out) // 0')
+  if [ "$t_in" -eq 0 ] && [ "$t_out" -eq 0 ]; then
+    printf 'Phase %s %s done (no LLM calls)' "$pid" "$name"
+    return 0
+  fi
+  usd="-"
+  [ "$idx" != "-" ] && usd=$(phase_usd "$idx")
+  printf 'Phase %s %s done - ~%s in / ~%s out tokens (%s, ~$%s)' \
+    "$pid" "$name" "$(format_tokens "$t_in")" "$(format_tokens "$t_out")" \
+    "${model:-unknown model}" "$usd"
+}
+
+# Printed after every status change. The card goes to stdout, and on every host
+# that collapses tool output the user never sees it - so the hint names the
+# host's own widget call, which the agent makes in its own turn where the user
+# does see it.
+tracker_next_hint() {
+  local pid="$1" status="$2" name mirror
+  [ "${TRACKER_QUIET:-0}" = "1" ] && return 0
+  name=$(load_state | jq -r --arg id "$pid" '(.phases[] | select(.id == $id) | .name) // ""')
+  case "$(host_kind)" in
+    claude) mirror="TaskUpdate(\"Phase $pid $name\", status=\"$status\")" ;;
+    codex) mirror="update_plan: set step \"Phase $pid $name\" to $status (send the FULL step list, it is not a delta)" ;;
+    *) mirror="no task widget on this host - reprint the card above inside your reply text" ;;
+  esac
+  printf '\n-- NEXT (required) --\n  %s\n' "$mirror"
+  case "$status" in
+    completed | failed)
+      printf '  narrate one line in outputLanguage: %s\n' "$(narration_line "$pid")"
+      ;;
+  esac
+}
+
+# Elapsed with an hours bucket. format_elapsed stops at minutes because the card
+# shows one phase at a time; a whole-run total reads as "102m" without this.
+format_span() {
+  local s="$1"
+  [ -z "$s" ] && { echo "-"; return; }
+  [ "$s" -lt 0 ] 2>/dev/null && { echo "-"; return; }
+  if [ "$s" -lt 3600 ]; then
+    format_elapsed "$s"
+  else
+    printf '%dh %dm' "$((s / 3600))" "$(((s % 3600) / 60))"
+  fi
+}
+
+# The end-of-run report: what the pipeline spent, phase by phase, and what it
+# has to say about phases it could not price. Printed by Phase 7 next to the
+# work summary, which covers what actually changed on disk.
+report() {
+  need_jq
+  local state task_id started
+  state="$(load_state)"
+  task_id=$(echo "$state" | jq -r '.task_id // "?"')
+  started=$(echo "$state" | jq -r '.started_at // ""')
+
+  printf '\n== Run report: %s ==\n' "$task_id"
+  printf '%-22s %-11s %10s   %-16s %-16s %8s\n' \
+    "Phase" "Status" "Elapsed" "Tokens in/out" "Model" "USD"
+  printf -- '%s\n' "----------------------------------------------------------------------------------"
+
+  local total_in=0 total_out=0 unpriced=""
+  local row
+  while IFS=$'\037' read -r idx id name status p_start p_end t_in t_out model no_llm; do
+    [ -n "$id" ] || continue
+    local secs="" span="-" toks="-" usd
+    if [ -n "$p_start" ]; then
+      local e_start e_end
+      e_start=$(iso_to_epoch "$p_start")
+      e_end=$([ -n "$p_end" ] && iso_to_epoch "$p_end" || now_epoch)
+      [ -n "$e_start" ] && [ -n "$e_end" ] && secs=$((e_end - e_start))
+      [ -n "$secs" ] && span=$(format_span "$secs")
+    fi
+    if [ "${t_in:-0}" -gt 0 ] || [ "${t_out:-0}" -gt 0 ]; then
+      toks="$(format_tokens "${t_in:-0}") / $(format_tokens "${t_out:-0}")"
+      total_in=$((total_in + t_in))
+      total_out=$((total_out + t_out))
+    elif [ "$status" = "completed" ] && [ "$no_llm" != "true" ]; then
+      unpriced="$unpriced $id"
+    fi
+    usd=$(phase_usd "$idx")
+    printf '%-22s %-11s %10s   %-16s %-16s %8s\n' \
+      "$id $name" "$status" "$span" "$toks" "${model:--}" "$usd"
+  done <<EOF
+$(echo "$state" | jq -r '
+  [.phases[]?] | to_entries[] |
+  [ (.key|tostring), .value.id, (.value.name // ""), (.value.status // "pending"),
+    (.value.started_at // ""), (.value.completed_at // ""),
+    ((.value.tokens_in // 0)|tostring), ((.value.tokens_out // 0)|tostring),
+    (.value.model // ""), ((.value.no_llm // false)|tostring) ] | join("\u001f")')
+EOF
+
+  printf -- '%s\n' "----------------------------------------------------------------------------------"
+  local run_secs="" run_span="-"
+  if [ -n "$started" ]; then
+    local e_run last
+    e_run=$(iso_to_epoch "$started")
+    last=$(echo "$state" | jq -r '[.phases[]?.completed_at // empty] | max // ""')
+    local e_last
+    e_last=$([ -n "$last" ] && iso_to_epoch "$last" || now_epoch)
+    [ -n "$e_run" ] && [ -n "$e_last" ] && run_secs=$((e_last - e_run))
+    [ -n "$run_secs" ] && run_span=$(format_span "$run_secs")
+  fi
+  printf '%-22s %-11s %10s   %-16s %-16s %8s\n' \
+    "Total" "" "$run_span" \
+    "$(format_tokens "$total_in") / $(format_tokens "$total_out")" "" "$(phase_usd total)"
+  if [ -n "$unpriced" ]; then
+    printf 'Cost unavailable  -  no tokens recorded for phase(s):%s\n' "$unpriced"
+  fi
+  printf '\n'
+}
+
 render() {
   need_jq
   local state
@@ -762,12 +925,49 @@ case "$ACTION" in
 
   update)
     need_jq
-    [ "$#" -ge 2 ] || { echo "update needs <phase_id> <status>" >&2; exit 64; }
-    PID="$1"; STATUS="$2"
+    [ "$#" -ge 2 ] || { echo "update needs <phase_id> <status> [--no-llm]" >&2; exit 64; }
+    PID="$1"; STATUS="$2"; shift 2
+    NO_LLM=0
+    while [ "$#" -gt 0 ]; do
+      case "$1" in
+        --no-llm) NO_LLM=1 ;;
+        *) echo "update: unknown option $1" >&2; exit 64 ;;
+      esac
+      shift
+    done
     case "$STATUS" in
       pending|in_progress|completed|failed|skipped) ;;
       *) echo "bad status: $STATUS" >&2; exit 64 ;;
     esac
+    # Accounting gate. A phase that ran LLM work and recorded nothing produces a
+    # completion line with nothing to say and a cost breakdown that reads as
+    # unavailable - which is exactly how per-phase token reporting went missing
+    # for months while every gate stayed green, because the gates lint the docs
+    # for the CALL and nothing checks that the call happened. Refusing the
+    # completion is recoverable (record, then re-run) and loses no work; a
+    # phase that genuinely made no LLM call says so with --no-llm.
+    if [ "$STATUS" = "completed" ] && [ "$NO_LLM" -eq 0 ]; then
+      case " ${TRACKER_LLM_PHASES:-1 2 3 4} " in
+        *" $PID "*)
+          RECORDED=$(load_state | jq -r --arg id "$PID" \
+            '(.phases[] | select(.id == $id) | (.tokens_in // 0) + (.tokens_out // 0)) // 0')
+          if [ "${RECORDED:-0}" -eq 0 ]; then
+            cat >&2 <<GATE
+update: phase $PID has no recorded spend, so its completion line would have
+nothing to report and the run report would price it as unavailable.
+
+Record it first, then re-run this update:
+  phase-tracker.sh model  $PID <model_name>
+  phase-tracker.sh tokens $PID <input_count> <output_count> [cached_count]
+
+If the phase genuinely made no LLM call, say so explicitly:
+  phase-tracker.sh update $PID completed --no-llm
+GATE
+            exit 3
+          fi
+          ;;
+      esac
+    fi
     NOW_ISO=$(date -u +"%Y-%m-%dT%H:%M:%SZ")
     acquire_state_lock
     state=$(load_state)
@@ -782,10 +982,14 @@ case "$ACTION" in
         else . end
       )
     ')
+    if [ "$NO_LLM" -eq 1 ]; then
+      new=$(echo "$new" | jq --arg id "$PID" '.phases |= map(if .id == $id then .no_llm = true else . end)')
+    fi
     save_state "$new"
     release_state_lock
     emit_otel_span "phase.update" "$PID" "$STATUS" "{\"status\": \"$STATUS\"}"
     render
+    tracker_next_hint "$PID" "$STATUS"
     usage_live_ping "$(basename "$TRACKER_DIR")" "$PID"
     ;;
 
@@ -893,7 +1097,11 @@ case "$ACTION" in
   model)
     need_jq
     [ "$#" -ge 2 ] || { echo "model needs <phase_id> <model_name>" >&2; exit 64; }
-    PID="$1"; MODEL="$2"
+    PID="$1"; MODEL="$(normalize_model "$2")"
+    if [ -f "$COST_TABLE" ] && ! jq -e --arg m "$MODEL" '.prices[$m]' "$COST_TABLE" >/dev/null 2>&1; then
+      echo "phase-tracker: '$2' is not a priced model  -  this phase will report USD as '-'." >&2
+      echo "  priced names: $(jq -r '.prices | keys | join(", ")' "$COST_TABLE" 2>/dev/null)" >&2
+    fi
     acquire_state_lock
     state=$(load_state)
     new=$(echo "$state" | jq --arg id "$PID" --arg m "$MODEL" '
@@ -948,6 +1156,40 @@ case "$ACTION" in
     fi
     ;;
 
+  tiles)
+    need_jq
+    tiles_state=$(load_state)
+    tiles_count=$(echo "$tiles_state" | jq '[.phases[]?] | length')
+    [ "${tiles_count:-0}" -gt 0 ] || {
+      echo "tiles: no phases registered  -  run 'add' for each phase first" >&2
+      exit 64
+    }
+    case "$(host_kind)" in
+      claude)
+        echo "REQUIRED  -  create one native tile per phase, in this exact order,"
+        echo "BEFORE any TaskUpdate. The widget renders by creation order, not by"
+        echo "phase number, so an out-of-order call scrambles the stack."
+        echo "$tiles_state" | jq -r '.phases[] | "  TaskCreate(subject: \"Phase \(.id) \(.name)\")"'
+        ;;
+      codex)
+        echo "REQUIRED  -  register the plan in ONE update_plan call with this step"
+        echo "list. update_plan takes the full list, not a delta, so every later"
+        echo "boundary resends it with one step's status changed."
+        echo "$tiles_state" | jq -c '{plan: [.phases[] | {step: "Phase \(.id) \(.name)", status: "pending"}]}'
+        ;;
+      *)
+        echo "Copilot CLI has no native task widget: the bordered card IS the widget."
+        echo "Reprint it inside your reply at every phase boundary  -  tool output is"
+        echo "collapsed, so a card left in stdout never reaches the user."
+        render
+        ;;
+    esac
+    ;;
+
+  report)
+    report
+    ;;
+
   render)
     render
     ;;

package/pipeline/scripts/pre-commit-check.sh

@@ -194,6 +194,7 @@ scan_file() {
           if (s ~ /^[0-9]+$/) next;        # pure digits
           if (s ~ /^[0-9a-f]+$/) next;     # lowercase hex (git sha / md5 / sha-*)
           if (s ~ /^[0-9A-F]+$/) next;     # uppercase hex
+          if (s ~ /^([A-Z][a-z]+)+[A-Z]*$/) next;  # CamelCase words only (label key / type name), never a credential
           delete freq;
           for (i = 1; i <= n; i++) { c = substr(s, i, 1); freq[c]++ }
           H = 0;

package/pipeline/scripts/uninstall.mjs

@@ -67,6 +67,7 @@ const HOME = process.env.HOME || process.env.USERPROFILE;
 export const PIPELINE_AGENT_FILES = [
   "android-architect.md",
   "backend-architect.md",
+  "bulk-reader.md",
   "code-reviewer.md",
   "dev-critic.md",
   "explorer.md",

package/pipeline/skills/.skill-manifest.json

@@ -1,7 +1,7 @@
 {
   "schemaVersion": "1.0.0",
-  "generatedAt": "2026-09-02T11:13:26Z",
-  "skillCount": 211,
+  "generatedAt": "2026-09-08T16:31:56Z",
+  "skillCount": 207,
   "entries": [
     {
       "path": "shared/core/apple-archive-compliance/SKILL.md",
@@ -43,22 +43,6 @@
       "path": "shared/core/multi-agent-design-check/SKILL.md",
       "sha256": "bdf5ef1b8e0b27eb216eed89382ced99552c4aadfb61bc9a4931e41e8316d92a"
     },
-    {
-      "path": "shared/core/multi-agent-dev-autopilot/SKILL.md",
-      "sha256": "42f8f5008a32202f7a51cec57d278baeaedca6d2ce70200dfce2fdd0eb62aaa4"
-    },
-    {
-      "path": "shared/core/multi-agent-dev-local-autopilot/SKILL.md",
-      "sha256": "257aa9d8ff48b76c07809d675f139087522edec598d5fb5bb0fcbd17e2857aac"
-    },
-    {
-      "path": "shared/core/multi-agent-dev-local/SKILL.md",
-      "sha256": "ba7e75447935e1c7bfc77bbbf8a0affe9cc2f631c2acea091de1b68d4bb34964"
-    },
-    {
-      "path": "shared/core/multi-agent-dev/SKILL.md",
-      "sha256": "4ac1401ebdd6d8ff21b85d11b1dffbd8399046348693eca758e6877a2ae07184"
-    },
     {
       "path": "shared/core/multi-agent-diff-explain/SKILL.md",
       "sha256": "7cb37a224e61d66e350202ecfc5fbc6c33da3fe0718504a093602f5cde5b648f"
@@ -81,11 +65,11 @@
     },
     {
       "path": "shared/core/multi-agent-help/SKILL.md",
-      "sha256": "5e4866b8ee203113b5db24e2d275323ae053254bbf677ad7f2f1e5c153fe1903"
+      "sha256": "87d45a04c782b03569b0ab9ee54da8414cbadd5a2a82a7701f38dc8a3ee94627"
     },
     {
       "path": "shared/core/multi-agent-ios-coding-standard/SKILL.md",
-      "sha256": "447a06d1a15928bf4569ee95d67332194b0db48a383be67a50eb436377d71019"
+      "sha256": "15a8d583cb7c772d5223c66c158062f25fdff4c41c6d353b7b70590a441dea38"
     },
     {
       "path": "shared/core/multi-agent-issue/SKILL.md",
@@ -137,7 +121,7 @@
     },
     {
       "path": "shared/core/multi-agent-resume-local/SKILL.md",
-      "sha256": "5b53ed0f6726082d0a4fb63f407578cd964df2f66ae3cc12f3f47fa3754fa50c"
+      "sha256": "a4bedf949ab648d977dd1409339d08af204341f555d69f71f58c1a54dffaff6d"
     },
     {
       "path": "shared/core/multi-agent-resume/SKILL.md",
@@ -193,11 +177,11 @@
     },
     {
       "path": "shared/core/multi-agent-store-ready/SKILL.md",
-      "sha256": "7ee313ed5dd0cb5ecd94ab68f8079323581a9337939bf60abc367163cf89fc57"
+      "sha256": "0060d1d7a8ec1e8cce30975f125faf46e9e384d23100ae46e82c9f194a0bf9fa"
     },
     {
       "path": "shared/core/multi-agent-sync/SKILL.md",
-      "sha256": "dc6f675aacbedc5c1516cbd7c624ee9cb17bb7c0136ed7effb66fe3cb401101d"
+      "sha256": "c0e2c22d0f6da9a60f5552d6f47db0490a355ea4ce08722716470e03535e5418"
     },
     {
       "path": "shared/core/multi-agent-test-accessibility/SKILL.md",
@@ -321,7 +305,7 @@
     },
     {
       "path": "shared/external/backlog/SKILL.md",
-      "sha256": "662618cb69086184e62d037d6589b22a9c53dc9c91f1381475986b4e62cb07f6"
+      "sha256": "d6602ec8f7bf70a305f4590ace8e8b92e28b165ffda2efb2fd95c8c27d541e23"
     },
     {
       "path": "shared/external/callkit-voip/SKILL.md",

package/pipeline/skills/.skills-index.json

@@ -1,6 +1,6 @@
 {
   "schemaVersion": "1.0.0",
-  "skillCount": 211,
+  "skillCount": 207,
   "entries": [
     {
       "name": "accessibility-compliance-accessibility-audit",
@@ -1036,50 +1036,6 @@
       "triggerPaths": [],
       "relativePath": "shared/core/multi-agent-design-check/SKILL.md"
     },
-    {
-      "name": "multi-agent-dev",
-      "description": "Removed in v16.0.0. Depth is a question the run asks, not a command name: start /multi-agent and answer Short. Invoke only to see that redirect.",
-      "platform": null,
-      "group": "core",
-      "plugin": null,
-      "invokeAs": "multi-agent-dev",
-      "triggerKeywords": [],
-      "triggerPaths": [],
-      "relativePath": "shared/core/multi-agent-dev/SKILL.md"
-    },
-    {
-      "name": "multi-agent-dev-autopilot",
-      "description": "Removed in v16.0.0 and not replaced: an unsupervised run is always the full pipeline now. Invoke only to be pointed at /multi-agent:autopilot or the interactive picker.",
-      "platform": null,
-      "group": "core",
-      "plugin": null,
-      "invokeAs": "multi-agent-dev-autopilot",
-      "triggerKeywords": [],
-      "triggerPaths": [],
-      "relativePath": "shared/core/multi-agent-dev-autopilot/SKILL.md"
-    },
-    {
-      "name": "multi-agent-dev-local",
-      "description": "Removed in v16.0.0. Its worktree-free twin is /multi-agent:local, which asks the same depth question; answer Short there. Invoke only to see that redirect.",
-      "platform": null,
-      "group": "core",
-      "plugin": null,
-      "invokeAs": "multi-agent-dev-local",
-      "triggerKeywords": [],
-      "triggerPaths": [],
-      "relativePath": "shared/core/multi-agent-dev-local/SKILL.md"
-    },
-    {
-      "name": "multi-agent-dev-local-autopilot",
-      "description": "Retired alongside its worktree twin in v16.0.0, with nothing standing in for it. Invoke only to be pointed at /multi-agent:local-autopilot or the interactive picker.",
-      "platform": null,
-      "group": "core",
-      "plugin": null,
-      "invokeAs": "multi-agent-dev-local-autopilot",
-      "triggerKeywords": [],
-      "triggerPaths": [],
-      "relativePath": "shared/core/multi-agent-dev-local-autopilot/SKILL.md"
-    },
     {
       "name": "multi-agent-diff-explain",
       "description": "Map Phase 4 triage findings to branch diff lines. Read-only post-hoc command, used after review to answer 'which finding lines up with which code change'. Use when a review finding has to be traced to the exact lines that caused it.",
@@ -1148,7 +1104,7 @@
     },
     {
       "name": "multi-agent-ios-coding-standard",
-      "description": "Audit an iOS module against the shared coding-standard registry (99 stable-ID rules), produce a remediation plan, then hand off to dev/dev-local. Use for a standards pass on a module, or when a review needs rule IDs rather than opinions.",
+      "description": "Audit an iOS module against the shared coding-standard registry (99 stable-ID rules), produce a remediation plan, then hand off to /multi-agent or :local. Use for a standards pass on a module, or when a review needs rule IDs rather than opinions.",
       "platform": null,
       "group": "core",
       "plugin": null,

package/pipeline/skills/shared/README.md

@@ -2,11 +2,11 @@
 
 Single source of truth for skills delivered to both Claude Code (`~/.claude/skills/`) and Copilot CLI (`~/.copilot/skills/`) by the installer.
 
-**Total:** 211 skills (58 core + 153 external). Auto-generated by `scripts/gen-skills-index.mjs`  -  do not edit by hand.
+**Total:** 207 skills (54 core + 153 external). Auto-generated by `scripts/gen-skills-index.mjs`  -  do not edit by hand.
 
 ## Directory layout
 
-- **`core/`**  -  58 `multi-agent*` orchestration skills that are pipeline-critical. Edits here are core-code changes.
+- **`core/`**  -  54 `multi-agent*` orchestration skills that are pipeline-critical. Edits here are core-code changes.
 - **`external/`**  -  153 iOS / Android / generic skills imported from the upstream skill library. Mirrors of third-party guidance.
 - Install destinations (ADR-0009): Claude Code gets NO local copy of `external/` - it loads those skills from the `multi-agent-plugins` marketplace, namespaced (`ai-<stack>-toolkit:<name>`); only the two compliance catalogs from `core/` land in `~/.claude/skills/`. Copilot CLI and Codex CLI receive a flat copy filtered to the enabled stacks. `external/` remains the single authoring source that `build-stack-plugins.mjs` publishes from.
 
@@ -14,7 +14,7 @@ Source layout is logical grouping only  -  skill discovery at runtime is unchang
 
 ## Categories
 
-- [Pipeline Orchestration](#pipeline-orchestration)  -  58
+- [Pipeline Orchestration](#pipeline-orchestration)  -  54
 - [iOS / Apple Ecosystem](#ios-apple-ecosystem)  -  90
 - [Android / Kotlin](#android-kotlin)  -  13
 - [Web](#web)  -  10
@@ -36,17 +36,13 @@ Source layout is logical grouping only  -  skill discovery at runtime is unchang
 | [`multi-agent-complaint-analysis`](./core/multi-agent-complaint-analysis/) | `core` | Customer-complaint triage. Ingests complaints (paste, csv/xlsx/txt/json file, Jira issue, Confluence URL), fetches Graylog evidence per trx/ |
 | [`multi-agent-create-jira`](./core/multi-agent-create-jira/) | `core` | Create a standards-compliant Jira issue (Task / Bug / Story): asks the type, mines project conventions, drafts from a standard template with |
 | [`multi-agent-design-check`](./core/multi-agent-design-check/) | `core` | Mock-mode vs Figma design audit (iOS / Android, local-only). Pick repo + module, gate on mock support, enumerate every state driver into a c |
-| [`multi-agent-dev`](./core/multi-agent-dev/) | `core` | Removed in v16.0.0. Depth is a question the run asks, not a command name: start /multi-agent and answer Short. Invoke only to see that redir |
-| [`multi-agent-dev-autopilot`](./core/multi-agent-dev-autopilot/) | `core` | Removed in v16.0.0 and not replaced: an unsupervised run is always the full pipeline now. Invoke only to be pointed at /multi-agent:autopilo |
-| [`multi-agent-dev-local`](./core/multi-agent-dev-local/) | `core` | Removed in v16.0.0. Its worktree-free twin is /multi-agent:local, which asks the same depth question; answer Short there. Invoke only to see |
-| [`multi-agent-dev-local-autopilot`](./core/multi-agent-dev-local-autopilot/) | `core` | Retired alongside its worktree twin in v16.0.0, with nothing standing in for it. Invoke only to be pointed at /multi-agent:local-autopilot o |
 | [`multi-agent-diff-explain`](./core/multi-agent-diff-explain/) | `core` | Map Phase 4 triage findings to branch diff lines. Read-only post-hoc command, used after review to answer 'which finding lines up with which |
 | [`multi-agent-feedback`](./core/multi-agent-feedback/) | `core` | Send one message to the maintainer: a bug, an idea or a question. Only the text you type is sent - no logs, no repo names, no paths. Shows t |
 | [`multi-agent-forget`](./core/multi-agent-forget/) | `core` | Remove a saved /multi-agent routine (created by /multi-agent:save): deletes its local-only command and its registry entry. Asks which one an |
 | [`multi-agent-garbage-collect`](./core/multi-agent-garbage-collect/) | `core` | Sweep leftover /tmp scratch (picker state, review diffs, channel payloads, analysis drafts) from past runs. Dry-run first; confirms before d |
 | [`multi-agent-graph`](./core/multi-agent-graph/) | `core` | Build and query this repo's code graph: a deterministic, LLM-free map of symbols, imports and references used to narrow Phase 1's Explore sc |
 | [`multi-agent-help`](./core/multi-agent-help/) | `core` | Multi-agent pipeline usage guide  -  renders in EN or TR per prefs.global.outputLanguage (falls back to promptLanguage for backward compatib |
-| [`multi-agent-ios-coding-standard`](./core/multi-agent-ios-coding-standard/) | `core` | Audit an iOS module against the shared coding-standard registry (99 stable-ID rules), produce a remediation plan, then hand off to dev/dev-l |
+| [`multi-agent-ios-coding-standard`](./core/multi-agent-ios-coding-standard/) | `core` | Audit an iOS module against the shared coding-standard registry (99 stable-ID rules), produce a remediation plan, then hand off to /multi-ag |
 | [`multi-agent-issue`](./core/multi-agent-issue/) | `core` | List unassigned GitHub issues, pick one, auto-assign, and launch the multi-agent pipeline. Use when a GitHub issue should be picked up and s |
 | [`multi-agent-jira`](./core/multi-agent-jira/) | `core` | List open Jira issues, pick one, and launch the multi-agent pipeline. Use when a Jira issue should be picked up and started without knowing  |
 | [`multi-agent-kill`](./core/multi-agent-kill/) | `core` | Stop the given task, then remove its worktree and branch. Asks for confirmation. Use when a running or stuck task should be stopped and its  |

package/pipeline/skills/shared/core/multi-agent-help/SKILL.md

@@ -87,10 +87,6 @@ Modes:
 Dedicated dash commands (Copilot):
   multi-agent-local, multi-agent-autopilot, multi-agent-local-autopilot
 
-Removed in v16.0.0: multi-agent-dev -> multi-agent + Short;
-  multi-agent-dev-local -> multi-agent-local + Short; the two dev autopilot
-  names have no equivalent (fast-plus-unattended no longer exists).
-
 ------------------------------------------------------------
 
 Utility Commands (dash form on Copilot, colon form on Claude Code):
@@ -264,10 +260,6 @@ Modlar:
 Dedicated dash komutlar (Copilot):
   multi-agent-local, multi-agent-autopilot, multi-agent-local-autopilot
 
-v16.0.0'da kaldırıldı: multi-agent-dev -> multi-agent + Kısa;
-  multi-agent-dev-local -> multi-agent-local + Kısa; iki dev autopilot adının
-  birebir karşılığı yok (hızlı+gözetimsiz artık yok).
-
 ------------------------------------------------------------
 
 Utility Komutları:

package/pipeline/skills/shared/core/multi-agent-ios-coding-standard/SKILL.md

@@ -1,7 +1,7 @@
 ---
 name: multi-agent-ios-coding-standard
 language: en
-description: "Audit an iOS module against the shared coding-standard registry (99 stable-ID rules), produce a remediation plan, then hand off to dev/dev-local. Use for a standards pass on a module, or when a review needs rule IDs rather than opinions."
+description: "Audit an iOS module against the shared coding-standard registry (99 stable-ID rules), produce a remediation plan, then hand off to /multi-agent or :local. Use for a standards pass on a module, or when a review needs rule IDs rather than opinions."
 user-invocable: true
 argument-hint: "[module name or path]"
 ---

package/pipeline/skills/shared/core/multi-agent-resume-local/SKILL.md

@@ -23,7 +23,7 @@ Phases 1-3 (Analysis / Planning / Dev) are skipped by design  -  the branch's lo
 
 ## When to use it
 
-- After `dev-local` / `local` (which skip Review + Test) to run the full quality tail on the same branch.
+- After a `:local` run answered Short (which skips Review + Test) to run the full quality tail on the same branch.
 - After hand-coding / hand-testing a change, to get review + build/test + PR + Jira write-up without re-running dev.
 
 ## When NOT to use it