refactor: batch-simplify sweep, waves 4-12 + deferred-work pass

README.md

@@ -116,7 +116,12 @@ genuinely carry no suite are recorded, with the CI lane that does cover them, in
 [`scripts/affected-tests-no-suite.txt`](scripts/affected-tests-no-suite.txt);
 `--allow-unmapped` is the escape hatch for everything else. That list is for
 prose and manifests, never for code: a source file with no coverage is supposed
-to fail here.
+to fail here. A **deletion** is the one exception: a changed path that no longer
+exists and that nothing claims is reported as a visible `deleted:` note instead
+of the error, because there is no content left to cover. A deletion that a
+surviving suite still names (a co-located test left behind, a suite that
+references the dead path) keeps selecting those suites, which are exactly what
+fails loudly if the deletion broke something.
 
 The runner is deliberately sequential: parallelising it measured sublinear
 (the suites are spawn-bound), and several guardrails suites assert wall-clock

package.json

@@ -13,6 +13,6 @@
     "markdownlint-cli2": "0.23.2"
   },
   "allowScripts": {
-    "@anthropic-ai/claude-code@2.1.209": true
+    "@anthropic-ai/claude-code@2.1.246": true
   }
 }

plugins/claude-config/.claude-plugin/plugin.json

@@ -1,7 +1,7 @@
 {
   "$schema": "https://json.schemastore.org/claude-code-plugin-manifest.json",
   "name": "claude-config",
-  "version": "0.40.19",
+  "version": "0.40.21",
   "description": "Nine configuration-health skills (plus setup) for a repo's Claude Code configuration: audit (settings.json / .mcp.json / hooks / plugins / permissions drift), audit-automation-gaps (evidence-gated verdicts on automation gaps), audit-permission-grants (allow-rule / allowed-tools grants for auto-mode durability and portability), audit-permission-state (the permission rules actually in effect \u2014 every settings scope merged with per-rule provenance, what auto mode drops on entry, config written where nothing reads it, and which managed intents are enforced versus loosenable), draft-auto-mode-rules (interview and draft a paste-ready autoMode classifier block; prints only, never writes), audit-instructions (locally-owned instruction surfaces vs current model capability \u2014 proposes removals/rewrites of instructions the model no longer needs, and detects cross-surface instruction conflicts), audit-prompting-postures (the additive lane \u2014 posture guidance the prompting guide says a component's purpose needs but the component does not carry), audit-pass (one coordinated, ordered, resumable pass over a named target \u2014 three-scope inventory, run-time-derived exclusion set, stable finding identity, suppression memory, resume, one human gate \u2014 delegating every check to the plugin that owns it), and unhobble (the empirical bare-baseline experiment: reversibly strip a repo's standing instructions, log real stumbles against the current model, re-add only what evidence earns).",
   "author": {
     "name": "Melodic Software",

plugins/claude-config/CHANGELOG.md

@@ -3,6 +3,28 @@
 All notable changes to the `claude-config` plugin are documented here. Format follows
 [Keep a Changelog](https://keepachangelog.com/en/1.1.0/); this plugin uses semantic versioning.
 
+## [0.40.21]
+
+### Changed
+
+- **`audit-permission-grants`: `permission-rule-check.test.sh` renumbers a colliding block of case
+  labels.** The regression block after Case 11 reused labels 8b through 8g that an earlier block
+  already carried, so two different cases shared each label. The later block is now Cases 12
+  through 13b, grouped by the issue each case pins. Comment-only: no assertion, variable, or
+  fixture changed, and the suite passes unchanged at 139 checks.
+
+## [0.40.20]
+
+### Changed
+
+- **Two documentation-accuracy fixes and one dead-guard removal in skill
+  scripts.** `audit-pass/scripts/run-state.sh`'s header usage block regains the
+  `--plugin-data` and `[--epoch <n>]` arguments its own `usage()` already
+  documents (comment-only; verified against the negative-test sed targets);
+  `audit-instructions/scripts/conflict-scan.test.sh` drops a prerequisite guard
+  for `grep`, a tool neither the suite nor the script under test invokes (the
+  awk guard stays; skip-discipline gates re-run clean).
+
 ## [0.40.19]
 
 ### Changed

plugins/claude-config/skills/audit-instructions/scripts/conflict-scan.test.sh

@@ -41,10 +41,6 @@ assert_not_contains() {
   esac
 }
 
-if ! command -v grep >/dev/null 2>&1; then
-  echo "SKIP: grep not installed" >&2
-  exit 0
-fi
 if ! command -v awk >/dev/null 2>&1; then
   echo "SKIP: awk not installed" >&2
   exit 0

plugins/claude-config/skills/audit-pass/scripts/run-state.sh

@@ -55,12 +55,12 @@
 #
 # Usage:
 #   run-state.sh paths          --plugin-data <dir> --run-id <id> [--root <path>]
-#   run-state.sh lease acquire  --run-dir <dir> --run-id <id> [--stale-after <s>]
-#                               [--skew-grace <s>] [--epoch <n>]
+#   run-state.sh lease acquire  --run-dir <dir> --run-id <id> --plugin-data <dir>
+#                               [--stale-after <s>] [--skew-grace <s>] [--epoch <n>]
 #   run-state.sh lease heartbeat --run-dir <dir>
 #   run-state.sh lease release   --run-dir <dir>
 #   run-state.sh lease classify  --run-dir <dir>
-#   run-state.sh partial append  --run-dir <dir> --record <json-line>
+#   run-state.sh partial append  --run-dir <dir> --record <json-line> [--epoch <n>]
 #
 # `--plugin-data` is required because `${CLAUDE_PLUGIN_DATA}` is NOT in the Bash
 # tool's environment (plugins reference: the three placeholders are exported to

plugins/claude-config/skills/audit-permission-grants/scripts/permission-rule-check.test.sh

@@ -563,7 +563,7 @@ assert_contains "refusal names the sanctioned variable as the fix" \
     -u CLAUDE_PROJECT_DIR GIT_CEILING_DIRECTORIES="$TEST_TMPDIR" bash "$SCRIPT" 2>&1)" \
   "PERMISSION_HYGIENE_SCAN_ROOT"
 
-# --- Case 8b: #2282 — full-rule reporting, `//` is NOT exempt, P1 pinned npm view
+# --- Case 12: #2282 — full-rule reporting, `//` is NOT exempt, P1 pinned npm view
 #
 # `//` is the ABSOLUTE anchor, not a portable one. permissions.md's own table row is
 # `//path` = "Absolute path from filesystem root", with `Read(//Users/<name>/secrets/**)`
@@ -582,15 +582,15 @@ assert_contains "P2 reports the full offending Bash rule" "$OUT_2282" "Bash(${PO
 assert_not_contains "fully-pinned npm view rule is not flagged as P1" "$OUT_2282" "npm view ctx7 version"
 assert_eq "both machine-path rules flagged, npm view not" "2" "$(run "$D8B" --count)"
 
-# --- Case 8c: the genuinely portable anchors stay exempt ----------------------
+# --- Case 12b: the genuinely portable anchors stay exempt ---------------------
 # The distinction the fix turns on: `~/` and `${CLAUDE_PROJECT_DIR}/` supply the
 # user/project segment at resolution time; `//` does not.
 D8C="$TEST_TMPDIR/issue-2282-portable"
 mkdir -p "$D8C/.claude"
 jq -n '{permissions:{allow:["Read(~/Documents/*.pdf)","Bash(${CLAUDE_PROJECT_DIR}/scripts/x.sh:*)"]}}' >"$D8C/.claude/settings.json"
 assert_eq "portable anchors produce no P2 finding" "0" "$(run "$D8C" --count)"
 
-# --- Case 8d: P2 reach is the open tool grammar, not five hardcoded names ------
+# --- Case 12c: P2 reach is the open tool grammar, not five hardcoded names -----
 # A hardcoded machine path leaks a username whatever tool the rule names. An
 # enumerated (Read|Edit|Write|Bash|PowerShell) list silently stopped flagging these;
 # `Agent` in particular is indefensible, since this script has a dedicated
@@ -604,15 +604,15 @@ assert_contains "P2 sees Glob rules" "$OUT_TOOLS" "Glob(${GLOB_MP})"
 assert_contains "P2 sees NotebookEdit rules" "$OUT_TOOLS" "NotebookEdit(${NB_MP})"
 assert_contains "P2 sees MCP tool rules" "$OUT_TOOLS" "mcp__srv__tool(${MCP_MP})"
 
-# --- Case 8e: a `//` prefix does not launder a path later in the same rule -----
+# --- Case 12d: a `//` prefix does not launder a path later in the same rule ----
 # Regression guard for a substring carve-out (`$m == *"(//"*`) that passed any rule
 # whose payload merely began with `//`, leaving the rest unexamined.
 D8E="$TEST_TMPDIR/issue-2282-traversal"
 mkdir -p "$D8E/.claude"
 jq -n --arg l "Read(${LAUNDER_MP})" '{permissions:{allow:[$l]}}' >"$D8E/.claude/settings.json"
 assert_eq "a // prefix does not exempt a user home later in the rule" "1" "$(run "$D8E" --count)"
 
-# --- Case 8f: #2397 A12 — tilde-user Bash paths leak a username ----------------
+# --- Case 13: #2397 A12 — tilde-user Bash paths leak a username ----------------
 D8F="$TEST_TMPDIR/issue-2397-tilde-user"
 mkdir -p "$D8F/.claude"
 jq -n '{permissions:{allow:["Bash(~kyle/scripts/x.sh:*)"]}}' >"$D8F/.claude/settings.json"
@@ -627,7 +627,7 @@ jq -n '{permissions:{allow:["Bash(curl https://example.com/~alice/index.html)"]}
   >"$D8F_URL/.claude/settings.json"
 assert_eq "URL user-directory segment is not flagged as tilde-user path" "0" "$(run "$D8F_URL" --count)"
 
-# --- Case 8g: #2397 A7b — inert substitution tokens in allowed-tools ----------
+# --- Case 13b: #2397 A7b — inert substitution tokens in allowed-tools ---------
 D8G="$TEST_TMPDIR/issue-2397-inert"
 mkdir -p "$D8G/.claude/skills/demo"
 cat >"$D8G/.claude/skills/demo/SKILL.md" <<'EOF'

plugins/claude-memory/.claude-plugin/plugin.json

@@ -1,7 +1,7 @@
 {
   "$schema": "https://json.schemastore.org/claude-code-plugin-manifest.json",
   "name": "claude-memory",
-  "version": "0.11.8",
+  "version": "0.11.9",
   "description": "Keeps a repo's Claude Code memory layer healthy and under your control, against criteria derived from official Claude Code documentation. The audit skill checks the instruction/memory layer (CLAUDE.md, CLAUDE.local.md, .claude/rules/, auto-memory) with a deterministic script-backed spine plus judgment-tier checks. The stateless skill inspects, disables, and (confirm-gated) purges Claude-written auto memory across all settings scopes.",
   "author": {
     "name": "Melodic Software",

plugins/claude-memory/CHANGELOG.md

@@ -3,6 +3,16 @@
 All notable changes to the `claude-memory` plugin are documented here. Format follows
 [Keep a Changelog](https://keepachangelog.com/en/1.1.0/); this plugin uses semantic versioning.
 
+## [0.11.9]
+
+### Changed
+
+- **`audit/scripts/memory-dir-stats.test.sh` fixture style normalized.** The
+  one fixture block that embedded literal newlines inside printf quotes (a
+  #2042 leftover) now uses the `\n`-escape form every other fixture in the
+  file uses; the produced fixture bytes are sha256-identical and all 64
+  checks pass.
+
 ## [0.11.8]
 
 ### Changed

plugins/claude-memory/skills/audit/scripts/memory-dir-stats.test.sh

@@ -184,13 +184,9 @@ assert_eq "a heavy one-line runaway counts every line" "$(raw_lines)" "$(run "$H
 # as octal escapes so the file stays ASCII and no tool can re-encode the fixture.
 WIDE=$(for ((i = 0; i < 600; i++)); do printf '\303\251'; done)
 {
-  printf -- '---
-'
-  printf 'Note: %s
-' "$WIDE"
-  printf -- '---
-body
-'
+  printf -- '---\n'
+  printf 'Note: %s\n' "$WIDE"
+  printf -- '---\nbody\n'
 } >"$M3/MEMORY.md"
 assert_eq "a multibyte heavy line is bounded by weight in bytes" "$(raw_bytes)" "$(run "$H3" --memory-bytes)"
 

plugins/claude-ops/.claude-plugin/plugin.json

@@ -1,7 +1,7 @@
 {
   "$schema": "https://json.schemastore.org/claude-code-plugin-manifest.json",
   "name": "claude-ops",
-  "version": "0.38.20",
+  "version": "0.38.21",
   "description": "Claude Code operations toolkit. Twelve skills: audit-skill-visibility (audit whether each installed skill is actually VISIBLE to the model, and diagnose why most of a fleet never gets used \u2014 a skill is invisible when its description is dropped by Claude Code's skill-listing context budget, which drops descriptions least-invoked-first so an unused skill loses the keywords that would let it be matched, from skills genuinely not wanted, from skills the run cannot observe at all; computes whether the listing overflows from documented settings, and withholds every cold verdict the data cannot support rather than reporting absence of data as absence of use), inventory (read-only enumeration of the complete invocable surface \u2014 every built-in CLI command with aliases and hidden/gated status, every bundled skill, and every component of every installed plugin across all marketplaces; reads the shipped binary because upstream publishes no built-in command list, and carries an integrity verdict so a drifted build reports counts as floors rather than silently short totals), audit-install-state (read-only audit of the machine-scope ~/.claude installation directory and ~/.claude.json \u2014 full inventory split into an authored surface and rolled-up bulk trees, product-managed retention vs genuinely unmanaged state, filename-scheme resolution before any process-liveness check, and deliberate/mid-experiment detection; reports, never deletes), audit-performance (read-only slowness-diagnostic capture run at the moment the machine or a session feels slow: CLI version, retention-sweep health including the silent unparsable-settings pause, a timed census walk of the install tree as a sweep-cost proxy, active-session and plugin-fleet counts, a process census, and the fan-out layer, which covers a load-labelled no-op spawn baseline, every hook that will fire bucketed per-tool-call versus per-turn with its invocation shape, the configured statusline, subagent concurrency and spawn-depth ceilings against documented defaults, whether running sessions predate the settings file they are judged by, and orphan attribution by parent liveness rather than age; read against a bundled known-performance-issues reference that also records the causes tested and cleared; separates the four documented suspects of accumulated state, version regression, component bloat, and per-spawn fan-out cost, and routes remediation out; reports, never mutates, and never executes a discovered hook or statusline command), audit-native-overlap (map native Claude Code surfaces \u2014 built-in CLI commands, bundled skills, plugin-backed built-ins, session-provided skills \u2014 against the current repo's plugin skills and agents, so a custom component never silently duplicates what Claude Code itself ships; bare invocation is a read-only overlap report carrying the extraction's integrity floors and a shared-listing-budget exposure section, verdicts are human-gated in a committed store rendered into a generated registry whose every row carries an observable recheck trigger, and only an explicit apply step bakes presence-gated native references into descriptions and Boundary sections), observability (read locally captured telemetry \u2014 OTEL store, collector, hook-event JSONL, ccusage \u2014 with trend reports and store pruning), known-issues (search known Claude product GitHub bugs, check service health, maintain a persistent tracked-issue registry), changelog (ingest Claude Code changelog entries and integrate them into the current repo), plugins (bring a machine's plugin fleet current on demand \u2014 marketplace refresh, effective-scope updates including in-repo project/local installs, new-plugin install per policy, scope-divergence detection and explicit convergence), morning-brief (read-only gh-based operator morning view \u2014 queue-label counts, merge-ready PRs, parked decisions with their RECOMMENDED lines, and loop-lane telemetry freshness), lanes (start/restart/stop/status loop lanes as named background Claude Code sessions seeded from canonical prompt files, with per-lane model/effort, a repo-pull + marketplace-refresh launch step, and a consume-restarts action \u2014 an OS-schedulable reader that relaunches stopped lanes whose telemetry carries a restart_request), and a re-runnable setup action that settles where the known-issues registry lives. Plus a family of eight advisory *-audit hooks (API errors, config changes, instruction loads, permission denials, pre-compaction, skill usage, tool failures, and unsurfaced hook failures \u2014 the last also warns the user via systemMessage, since a hook that fails to launch enforces nothing and Claude Code surfaces the failure to nobody) that emit the shared hook-telemetry envelope, and a reference sink that maps envelopes into the hook-events.jsonl the observability skill reads.",
   "author": {
     "name": "Melodic Software",

plugins/claude-ops/CHANGELOG.md

@@ -3,6 +3,25 @@
 All notable changes to the `claude-ops` plugin are documented here. Format follows
 [Keep a Changelog](https://keepachangelog.com/en/1.1.0/); this plugin uses semantic versioning.
 
+## [0.38.21]
+
+### Changed
+
+- **One fewer spawn per Stop turn in `hook-failure-audit.sh`.** The hook now
+  short-circuits before its summary jq spawn when the grep pre-filter matched
+  nothing (the empty case previously produced `[]` and the same silent exit 0).
+  Adversarial payload probes byte-identical across all paths; suite 88/88.
+  (An analogous jq_fields consolidation in the two skill-usage hooks was
+  attempted, refuted by differential testing on pathological payloads, and
+  reverted — recorded in the sweep report rather than shipped.)
+- **restart-consumer.sh declares five formerly implicit globals `local` to
+  `process_lane`** (no post-return reader exists; full caller-graph and trap
+  audit) and drops two `${lock_rc:-0}` defaults dominated by an unconditional
+  assignment. `morning-brief.sh` declares its `read -ra` scratch array local;
+  two misplaced test assertions moved to the section whose fixture they read.
+- **install_state.py** renames the unused `os.walk` dirnames slot to
+  `_dirnames`, matching the file's own convention.
+
 ## [0.38.20]
 
 ### Changed

plugins/claude-ops/hooks/hook-failure-audit.sh

@@ -79,7 +79,9 @@ read_window() {
   fi
 }
 
-# grep is a cheap pre-filter only; the structural jq selection decides.
+# grep is a cheap pre-filter only; the structural jq selection decides. The
+# no-match common case exits on the pre-filter's emptiness, before paying for
+# the jq spawn (an empty stream produced the same silent exit via "[]").
 # `fromjson?` skips unparsable lines instead of aborting the stream.
 # Identity is the REGISTRATION, not the matcher name: several plugins register
 # on the same event+matcher (multiple PreToolUse:Bash guards exist in this very
@@ -130,7 +132,9 @@ read_window() {
 # in a narrower shape. The per-class counts below keep every class present in a
 # group visible, and the message flags are computed from those counts, never
 # from a single collapsed value.
-SUMMARY=$(read_window | grep -F '"hook_non_blocking_error"' |
+RECORDS=$(read_window | grep -F '"hook_non_blocking_error"')
+[[ -n "$RECORDS" ]] || exit 0
+SUMMARY=$(printf '%s' "$RECORDS" |
   jq -cRs '[
       split("\n")[] | fromjson?
       | select(.type? == "attachment") | .attachment

plugins/claude-ops/skills/audit-install-state/scripts/install_state.py

@@ -847,7 +847,7 @@ def walk_tree(
     exclude = exclude or set()
     rows: list[FileRow] = []
     errors: list[dict] = []
-    for dirpath, dirnames, filenames in os.walk(
+    for dirpath, _dirnames, filenames in os.walk(
         root, followlinks=False, onerror=errors.append
     ):
         for name in filenames:

plugins/claude-ops/skills/lanes/scripts/restart-consumer.sh

@@ -924,7 +924,8 @@ lane_selected() {
 
 process_lane() {
   local idx="$1" now="$2"
-  local lane repo issue marker bodies body state request reason used n i found=0
+  local lane repo issue marker instance bodies body state request req reason used n i found=0
+  local comment_marker sibling_request sibling_instance
   lane="$(lane_field "$idx" name)"
   [[ -n "$lane" ]] || return 0
   lane_selected "$lane" || return 0
@@ -1271,11 +1272,11 @@ main() {
     acquire_lock "$now" || lock_rc=$?
     # 2 = the lock store itself is unusable. Exiting 0 here would let an
     # unattended schedule log healthy ticks forever while processing nothing.
-    if ((${lock_rc:-0} == 2)); then
+    if ((lock_rc == 2)); then
       err "restart-consumer: the lock store is unusable — not a held lock; failing rather than reporting a skipped tick"
       exit 4
     fi
-    if ((${lock_rc:-0} != 0)); then
+    if ((lock_rc != 0)); then
       warn "another restart-consumer run holds the lock ($LOCK_DIR) — skipping this tick"
       info "restart-consumer: $ACTION on ${TARGET_REPO:-<fixtures>} at $(iso_utc "$now")"
       info "| lane | decision | detail |"