@boyingliu01/xp-gate 0.34.1 → 0.35.0

package/gate-3.sh

@@ -461,6 +461,52 @@ else
           GATE_3_DETAIL="mode=ratchet verifier=MISSING"
           exit 1
         fi
+        RATCHET_LIB="$GATE_DIR/lib/ccn-ratchet.sh"
+        [ -f "$RATCHET_LIB" ] || RATCHET_LIB="$(git rev-parse --show-toplevel 2>/dev/null)/githooks/lib/ccn-ratchet.sh"
+        # #617: grade exactly the population the baseline was written from.
+        #
+        # The comment below has always said so, but only as a comment: a repo whose
+        # baseline came from a narrower scan (a Java-only ccn-report, say) had its
+        # vendored minified JS graded anyway, and --csv-verified then disagreed with
+        # lizard on single-line code and fail-closed with "could not be
+        # span-verified". That BLOCK has no repair: the file has no baseline entry
+        # to ratchet, so there is nothing the user could add to clear it.
+        #
+        # The population filter runs BEFORE verification on purpose. Domain
+        # restriction is decidable from the path alone, so asking the expensive
+        # span re-derivation about a file that will not be graded anyway only
+        # manufactures failures for code the ratchet never claims to cover.
+        if [ -f "$RATCHET_LIB" ]; then
+          # shellcheck source=/dev/null
+          . "$RATCHET_LIB" 2>/dev/null || true
+          if declare -F ccn_file_in_population >/dev/null 2>&1; then
+            _cc_pop_kept=""
+            _cc_pop_skipped=""
+            while IFS= read -r _cc_row; do
+              [ -n "$_cc_row" ] || continue
+              _cc_row_path=$(printf '%s' "$_cc_row" | awk -F, '{print $1}')
+              if ccn_file_in_population ".ccn-baseline.json" "$_cc_row_path"; then
+                _cc_pop_kept="${_cc_pop_kept}${_cc_row}
+"
+              else
+                _cc_pop_skipped="${_cc_pop_skipped}${_cc_row_path}
+"
+              fi
+            done <<CC_POP_EOF
+$CC_CSV
+CC_POP_EOF
+            CC_CSV=$(printf '%s' "$_cc_pop_kept")
+            if [ -n "$_cc_pop_skipped" ]; then
+              _cc_pop_skip_n=$(printf '%s' "$_cc_pop_skipped" | grep -c . || true)
+              echo "ℹ️  Gate 3: ${_cc_pop_skip_n:-0} file(s) outside the baseline population — not graded, not ratcheted:"
+              printf '%s' "$_cc_pop_skipped" | sed 's/^/     /' | sed -n '1,5p'
+              if [ "${_cc_pop_skip_n:-0}" -gt 5 ]; then
+                echo "     ... and $((_cc_pop_skip_n - 5)) more"
+              fi
+              echo "     (declare or widen the population in .ccn-baseline.json to grade them)"
+            fi
+          fi
+        fi
         if [ -f "$CC_REPORT" ] && command -v node >/dev/null 2>&1; then
           _cc_verify_failed=0
           CC_CSV_VERIFIED=$(printf '%s\n' "$CC_CSV" | node "$CC_REPORT" --csv-verified batch --root "$(git rev-parse --show-toplevel 2>/dev/null)") || _cc_verify_failed=1
@@ -471,8 +517,6 @@ else
           fi
           CC_CSV="$CC_CSV_VERIFIED"
         fi
-        RATCHET_LIB="$GATE_DIR/lib/ccn-ratchet.sh"
-        [ -f "$RATCHET_LIB" ] || RATCHET_LIB="$(git rev-parse --show-toplevel 2>/dev/null)/githooks/lib/ccn-ratchet.sh"
         # Fail closed when the ratchet library is missing. Sourcing a nonexistent
         # file only prints "No such file or directory" and leaves
         # evaluate_ccn_ratchet undefined; the verdict then came back empty, every

package/gate-4.sh

@@ -172,9 +172,26 @@ else
         rm -f "$PRINCIPLES_STDERR"
 
         if [ "$PRINCIPLES_EXIT" -ge 2 ]; then
-          echo "⚠️  Warning: Principles checker execution failed"
-          echo "⏭️  SKIPPED - Principles check (execution issue)"
-          GATE_4_STATUS="SKIP"
+          # #661: exit>=2 means the checker did not RUN (crash, module not found,
+          # bad invocation). This used to record SKIP, and pre-commit treats SKIP
+          # as non-blocking, so a tool failure was absorbed into the green light --
+          # the reproduction printed a module-not-found stack trace and then
+          # "✅ PASSED - Principles checker (no errors found)".
+          #
+          # The distinction the gate must keep:
+          #   - tool NOT INSTALLED  -> SKIP is honest; nothing is being claimed.
+          #   - tool PRESENT, CRASH -> WARN. Not a pass (nothing was checked) and
+          #     not a silent skip (the user must know their code went unaudited).
+          #
+          # WARN rather than FAIL on purpose: the commit itself is not necessarily
+          # bad, the TOOL is broken, and blocking every commit until node/tsx is
+          # repaired would blame the change for an environment fault. What must
+          # never happen is the green light.
+          echo "⚠️  Principles checker did not run (exit ${PRINCIPLES_EXIT}) — this code is NOT verified."
+          echo "     Gate 4 cannot claim a pass for a check that never executed."
+          echo "     Fix the checker invocation above (or run 'npx xp-gate check principles .')."
+          echo "⏭️  WARN - Principles check (checker execution error)"
+          GATE_4_STATUS="WARN"
           # Consumed by the sourcing parent, not here (see #548).
           # shellcheck disable=SC2034
           GATE_4_REASON="checker_execution_error"

package/hooks/gate-3.sh

@@ -461,6 +461,52 @@ else
           GATE_3_DETAIL="mode=ratchet verifier=MISSING"
           exit 1
         fi
+        RATCHET_LIB="$GATE_DIR/lib/ccn-ratchet.sh"
+        [ -f "$RATCHET_LIB" ] || RATCHET_LIB="$(git rev-parse --show-toplevel 2>/dev/null)/githooks/lib/ccn-ratchet.sh"
+        # #617: grade exactly the population the baseline was written from.
+        #
+        # The comment below has always said so, but only as a comment: a repo whose
+        # baseline came from a narrower scan (a Java-only ccn-report, say) had its
+        # vendored minified JS graded anyway, and --csv-verified then disagreed with
+        # lizard on single-line code and fail-closed with "could not be
+        # span-verified". That BLOCK has no repair: the file has no baseline entry
+        # to ratchet, so there is nothing the user could add to clear it.
+        #
+        # The population filter runs BEFORE verification on purpose. Domain
+        # restriction is decidable from the path alone, so asking the expensive
+        # span re-derivation about a file that will not be graded anyway only
+        # manufactures failures for code the ratchet never claims to cover.
+        if [ -f "$RATCHET_LIB" ]; then
+          # shellcheck source=/dev/null
+          . "$RATCHET_LIB" 2>/dev/null || true
+          if declare -F ccn_file_in_population >/dev/null 2>&1; then
+            _cc_pop_kept=""
+            _cc_pop_skipped=""
+            while IFS= read -r _cc_row; do
+              [ -n "$_cc_row" ] || continue
+              _cc_row_path=$(printf '%s' "$_cc_row" | awk -F, '{print $1}')
+              if ccn_file_in_population ".ccn-baseline.json" "$_cc_row_path"; then
+                _cc_pop_kept="${_cc_pop_kept}${_cc_row}
+"
+              else
+                _cc_pop_skipped="${_cc_pop_skipped}${_cc_row_path}
+"
+              fi
+            done <<CC_POP_EOF
+$CC_CSV
+CC_POP_EOF
+            CC_CSV=$(printf '%s' "$_cc_pop_kept")
+            if [ -n "$_cc_pop_skipped" ]; then
+              _cc_pop_skip_n=$(printf '%s' "$_cc_pop_skipped" | grep -c . || true)
+              echo "ℹ️  Gate 3: ${_cc_pop_skip_n:-0} file(s) outside the baseline population — not graded, not ratcheted:"
+              printf '%s' "$_cc_pop_skipped" | sed 's/^/     /' | sed -n '1,5p'
+              if [ "${_cc_pop_skip_n:-0}" -gt 5 ]; then
+                echo "     ... and $((_cc_pop_skip_n - 5)) more"
+              fi
+              echo "     (declare or widen the population in .ccn-baseline.json to grade them)"
+            fi
+          fi
+        fi
         if [ -f "$CC_REPORT" ] && command -v node >/dev/null 2>&1; then
           _cc_verify_failed=0
           CC_CSV_VERIFIED=$(printf '%s\n' "$CC_CSV" | node "$CC_REPORT" --csv-verified batch --root "$(git rev-parse --show-toplevel 2>/dev/null)") || _cc_verify_failed=1
@@ -471,8 +517,6 @@ else
           fi
           CC_CSV="$CC_CSV_VERIFIED"
         fi
-        RATCHET_LIB="$GATE_DIR/lib/ccn-ratchet.sh"
-        [ -f "$RATCHET_LIB" ] || RATCHET_LIB="$(git rev-parse --show-toplevel 2>/dev/null)/githooks/lib/ccn-ratchet.sh"
         # Fail closed when the ratchet library is missing. Sourcing a nonexistent
         # file only prints "No such file or directory" and leaves
         # evaluate_ccn_ratchet undefined; the verdict then came back empty, every

package/hooks/gate-4.sh

@@ -172,9 +172,26 @@ else
         rm -f "$PRINCIPLES_STDERR"
 
         if [ "$PRINCIPLES_EXIT" -ge 2 ]; then
-          echo "⚠️  Warning: Principles checker execution failed"
-          echo "⏭️  SKIPPED - Principles check (execution issue)"
-          GATE_4_STATUS="SKIP"
+          # #661: exit>=2 means the checker did not RUN (crash, module not found,
+          # bad invocation). This used to record SKIP, and pre-commit treats SKIP
+          # as non-blocking, so a tool failure was absorbed into the green light --
+          # the reproduction printed a module-not-found stack trace and then
+          # "✅ PASSED - Principles checker (no errors found)".
+          #
+          # The distinction the gate must keep:
+          #   - tool NOT INSTALLED  -> SKIP is honest; nothing is being claimed.
+          #   - tool PRESENT, CRASH -> WARN. Not a pass (nothing was checked) and
+          #     not a silent skip (the user must know their code went unaudited).
+          #
+          # WARN rather than FAIL on purpose: the commit itself is not necessarily
+          # bad, the TOOL is broken, and blocking every commit until node/tsx is
+          # repaired would blame the change for an environment fault. What must
+          # never happen is the green light.
+          echo "⚠️  Principles checker did not run (exit ${PRINCIPLES_EXIT}) — this code is NOT verified."
+          echo "     Gate 4 cannot claim a pass for a check that never executed."
+          echo "     Fix the checker invocation above (or run 'npx xp-gate check principles .')."
+          echo "⏭️  WARN - Principles check (checker execution error)"
+          GATE_4_STATUS="WARN"
           # Consumed by the sourcing parent, not here (see #548).
           # shellcheck disable=SC2034
           GATE_4_REASON="checker_execution_error"

package/hooks/lib/ccn-ratchet.sh

@@ -11,6 +11,23 @@
 # improves we say so and suggest refreshing the baseline, because a stale
 # ceiling invites nobody to finish the job.
 #
+# ── Population (#617) ───────────────────────────────────────────────────────
+# A ratchet is only meaningful over the population it was recorded from. Gate 3
+# grades every extension it recognises; a repo whose baseline came from a
+# narrower scan then reports pre-existing code as "new", which is a false BLOCK
+# no repair can clear — the file has no baseline entry to fix.
+#
+# The baseline may therefore DECLARE its population:
+#
+#   { "population": { "roots": ["src/main/java"], "extensions": [".java"] },
+#     "entries": { "src/main/java/A.java::f": 4 } }
+#
+# A baseline without that key is LEGACY and is treated as unrestricted, which is
+# exactly the historical behaviour: every baseline already in the field has the
+# flat `{ "path::fn": ccn }` shape, and refusing to load it would break every
+# install on upgrade. Absent metadata must never be read as an empty population
+# -- that would silently grade nothing at all, the opposite failure.
+#
 # Key = "<relative/path>::<functionName>". Rename or move = new key = treated
 # as a new function. That is deliberate: fuzzy matching would hide debt moves.
 # `node scripts/ccn-report.cjs --write-baseline` regenerates the file after a
@@ -41,6 +58,98 @@
 # every row is read as a single field).
 CSV_FS=$'\x1f'
 
+# ---------------------------------------------------------------------------
+# #617: baseline population.
+#
+# Prints the declared population as `<root>|<ext>|<ext>...` on one line, or
+# NOTHING when the baseline carries no `population` key (legacy => unrestricted).
+# Printing nothing is deliberate: "no declaration" and "empty population" are
+# opposite meanings, and a caller that cannot tell them apart would either grade
+# everything (wrong for a Java-only repo) or nothing (catastrophic -- the ratchet
+# would report a clean run while comparing zero functions).
+# ---------------------------------------------------------------------------
+ccn_baseline_population() {
+  local _bl="${1:-.ccn-baseline.json}"
+  [ -f "$_bl" ] || return 0
+  command -v node >/dev/null 2>&1 || return 0
+  node -e '
+    const fs = require("fs");
+    let data;
+    try { data = JSON.parse(fs.readFileSync(process.argv[1], "utf8")); }
+    catch { process.exit(0); }
+    // Legacy shape: a flat map of "path::fn" -> ccn. No declaration.
+    if (!data || typeof data !== "object" || !data.population) process.exit(0);
+    const pop = data.population || {};
+    const roots = Array.isArray(pop.roots) ? pop.roots : [];
+    const exts = Array.isArray(pop.extensions) ? pop.extensions : [];
+    // A declared-but-empty population is not a legitimate declaration: it would
+    // excuse every file. Treat it as no declaration so the gate keeps grading.
+    if (roots.length === 0 && exts.length === 0) process.exit(0);
+    // One line: "<roots csv>|<exts csv>". Callers split on the single pipe.
+    process.stdout.write(roots.join(",") + "|" + exts.join(","));
+  ' "$_bl" 2>/dev/null
+  return 0
+}
+
+# ---------------------------------------------------------------------------
+# #617: is this file inside the baseline's declared population?
+#
+# Returns 0 when the file may be graded, 1 when it is outside the population.
+# A legacy baseline (no declaration) permits everything, preserving the
+# behaviour every existing install already relies on.
+#
+# Both axes are ANDed only when both are declared; a baseline declaring just
+# roots (or just extensions) narrows on that axis alone.
+# ---------------------------------------------------------------------------
+ccn_file_in_population() {
+  local _bl="${1:-.ccn-baseline.json}"
+  local _file="${2:-}"
+  [ -n "$_file" ] || return 1
+
+  local _pop
+  _pop=$(ccn_baseline_population "$_bl")
+  # No declaration => unrestricted (legacy baseline).
+  [ -n "$_pop" ] || return 0
+
+  local _roots="${_pop%%|*}"
+  local _exts="${_pop#*|}"
+  # Normalise separators so a Windows path and a POSIX path compare the same way.
+  # sed, not `tr '\' '/'`: a lone backslash is a portability trap in tr's SET
+  # argument (GNU tr warns "unescaped backslash at end of string" and MSYS tr
+  # behaviour differs), while the sed regex below is unambiguous everywhere.
+  _norm=$(printf '%s' "$_file" | sed 's|\\|/|g')
+
+  if [ -n "$_roots" ]; then
+    local _hit=0 _r
+    local _oldifs="$IFS"
+    IFS=','
+    for _r in $_roots; do
+      IFS="$_oldifs"
+      case "$_norm" in
+        "$_r"|"$_r"/*) _hit=1; break ;;
+      esac
+    done
+    IFS="$_oldifs"
+    [ "$_hit" -eq 1 ] || return 1
+  fi
+
+  if [ -n "$_exts" ]; then
+    local _match=0 _e
+    local _oldifs2="$IFS"
+    IFS=','
+    for _e in $_exts; do
+      IFS="$_oldifs2"
+      case "$_norm" in
+        *"$_e") _match=1; break ;;
+      esac
+    done
+    IFS="$_oldifs2"
+    [ "$_match" -eq 1 ] || return 1
+  fi
+
+  return 0
+}
+
 csv_to_unit_separated() {
   awk '
     function emit(line,   cur, i, ch, q, nf) {

package/lib/__tests__/doctor-timeout-partial.test.js

@@ -0,0 +1,79 @@
+/**
+ * @test REQ-629
+ * @intent `xp-gate doctor` is non-deterministic on Windows. Against an identical,
+ *         freshly-installed, fully healthy installation it printed
+ *         "All checks passed" on one run and "Timed out after 10s — some checks
+ *         incomplete" on the next two. The timeout is a deliberate safety net
+ *         (#348, "43s on Windows"); the defect is what happens when it fires.
+ *
+ *         `Promise.race` discarded the entire `diagnoseAsync()` result and
+ *         substituted a single placeholder check, so everything that had already
+ *         finished was thrown away. Consequence: doctor can report LESS
+ *         information the LONGER it runs — the opposite of what a diagnostic
+ *         should do — and a WARN from the timeout made the skills section vanish
+ *         entirely, so tooling that scrapes `Skill:.*Up to date` read 0/12.
+ *
+ *         The fix: reach a verdict at the deadline (never hang — #348 stays fixed)
+ *         but keep whatever finished. A diagnostic that says "these 9 checks
+ *         passed, 2 did not report in time" is strictly more useful than one that
+ *         says "timed out".
+ *
+ *         The tests below pin the merge semantics directly rather than waiting on
+ *         a real 10s timer: the behaviour that matters is "partial results are
+ *         preserved and reported as partial", not the clock.
+ * @covers AC-670-43, AC-670-44, AC-670-45
+ */
+'use strict';
+
+// describe/it/expect are vitest globals in this repo's config; requiring them from
+// 'vitest' fails because this file is CommonJS. (Hooks are not needed here.)
+
+/** A completed check, shaped like diagnoseAsync()'s output. */
+function check(name, status) {
+  return { name, status, detail: `${name} detail` };
+}
+
+describe('doctor keeps completed checks when the global timeout fires (#629)', () => {
+  it('AC-670-43: completed checks are merged with the timeout notice, not replaced', () => {
+    const { mergeTimedOutDiagnosis } = require('../doctor');
+    expect(typeof mergeTimedOutDiagnosis).toBe('function');
+
+    const completed = {
+      checks: [check('Hooks', 'PASS'), check('Adapters', 'PASS'), check('Skills', 'WARN')],
+      issues: 1,
+    };
+    const merged = mergeTimedOutDiagnosis(completed, ['Gate env', 'OpenCode TUI']);
+
+    // Nothing that finished may be dropped -- that is the whole defect.
+    for (const name of ['Hooks', 'Adapters', 'Skills']) {
+      expect(merged.checks.map((c) => c.name)).toContain(name);
+    }
+    // The deadline itself is still reported, and still not a pass.
+    const notice = merged.checks.find((c) => c.name === 'Diagnosis');
+    expect(notice).toBeDefined();
+    expect(notice.status).toBe('WARN');
+    // Named individually, so the reader knows which surface is unknown.
+    expect(notice.detail).toContain('Gate env');
+    expect(notice.detail).toContain('OpenCode TUI');
+    expect(merged.timedOut).toBe(true);
+    // A partial run is never clean.
+    expect(merged.issues).toBeGreaterThanOrEqual(1);
+  });
+
+  it('AC-670-44: an empty partial result still yields the honest timeout notice', () => {
+    const { mergeTimedOutDiagnosis } = require('../doctor');
+    const merged = mergeTimedOutDiagnosis({ checks: [], issues: 0 }, ['Gate env']);
+    expect(merged.checks).toHaveLength(1);
+    expect(merged.checks[0].name).toBe('Diagnosis');
+    expect(merged.timedOut).toBe(true);
+  });
+
+  it('AC-670-45: nothing that already reported is ever lost, and a real issue count survives', () => {
+    const { mergeTimedOutDiagnosis } = require('../doctor');
+    const completed = { checks: [check('Hooks', 'FAIL'), check('Templates', 'PASS')], issues: 4 };
+    const merged = mergeTimedOutDiagnosis(completed, []);
+    // Pre-existing issues are not reset by the timeout.
+    expect(merged.issues).toBeGreaterThanOrEqual(4);
+    expect(merged.checks.map((c) => c.name)).toEqual(expect.arrayContaining(['Hooks', 'Templates']));
+  });
+});

package/lib/doctor.js

@@ -80,6 +80,38 @@ const EXEC_TIMEOUT_MS = 1500;
  */
 const GLOBAL_DIAGNOSIS_TIMEOUT_MS = 10000;
 
+/**
+ * Merge a timed-out partial diagnosis into one honest verdict (#629).
+ *
+ * `doctor`'s global timeout (#348) exists so the command never hangs, but the
+ * original handler threw away the entire `diagnoseAsync()` result and substituted
+ * a single placeholder check. Against an identical healthy install that made the
+ * SAME command print "All checks passed" or "Timed out after 10s" depending on
+ * machine load, and a timed-out run reported 0 installed skills — doctor giving
+ * LESS information the LONGER it runs is the opposite of a diagnostic.
+ *
+ * Keeps every completed check, adds ONE Diagnosis notice naming what did not
+ * report, and never lets a partial run read as clean (#495).
+ *
+ * @param {{checks?: Array, issues?: number}} partial checks finished before the deadline
+ * @param {string[]} pending names of surfaces that did not report in time
+ */
+function mergeTimedOutDiagnosis(partial, pending) {
+  const completed = (partial && Array.isArray(partial.checks)) ? partial.checks : [];
+  const inherited = (partial && Number.isFinite(partial.issues)) ? partial.issues : 0;
+  const pendingNames = Array.isArray(pending) ? pending.filter(Boolean) : [];
+  const detail = pendingNames.length > 0
+    ? `Timed out after ${GLOBAL_DIAGNOSIS_TIMEOUT_MS / 1000}s — did not report in time: ${pendingNames.join(', ')}`
+    : `Timed out after ${GLOBAL_DIAGNOSIS_TIMEOUT_MS / 1000}s — some checks incomplete`;
+  return {
+    checks: [...completed, { name: 'Diagnosis', status: 'WARN', detail }],
+    // A partial run is never clean, and a real issue count is never reset.
+    issues: Math.max(inherited, 1),
+    timedOut: true,
+    completedCheckCount: completed.length,
+  };
+}
+
 /**
  * Fixed attribution text for the WorkBuddy-desktop sandbox restriction (#525).
  *
@@ -1605,9 +1637,17 @@ async function checkAdapters(checks, mode, gitDirPromise) {
  *   D) Env + CLI tools — fully independent, run IN PARALLEL
  *   E) TUI + Skills — independent, run IN PARALLEL
  */
-async function diagnoseAsync(refreshEnv = false) {
+async function diagnoseAsync(refreshEnv = false, progress = null) {
+  // #629: `progress` is a live view of the checks completed so far. The global
+  // timeout used to discard everything diagnoseAsync had already produced; with a
+  // shared sink, a deadline can report "these finished, those did not" instead of
+  // replacing the whole diagnosis with a placeholder.
   const checks = [];
   let issues = 0;
+  if (progress) {
+    progress.checks = checks;
+    progress.getIssues = () => issues;
+  }
 
   // Spawn capability probe FIRST (sync, no config dependency) — #525: it must
   // report even when the config early-exits below, and its WARN (environmental,
@@ -2869,10 +2909,26 @@ async function doctor(args) {
 
   // Global timeout wrapper — ensures doctor completes within GLOBAL_DIAGNOSIS_TIMEOUT_MS
   // even if network/subprocess checks stack up (fix #348: 43s on Windows).
-  const diagnosisPromise = diagnoseAsync(refreshEnv);
+  // #629: give diagnoseAsync a live sink so a deadline can keep what finished.
+  // Without it, `Promise.race` discarded the entire diagnosis -- so the SAME
+  // healthy install reported different results run to run, and a timed-out run
+  // lost the skills section entirely (scrapers read 0/12 installed skills).
+  const progress = { checks: [], getIssues: () => 0 };
+  const diagnosisPromise = diagnoseAsync(refreshEnv, progress);
   let diagnosisTimer = null;
   const timeoutPromise = new Promise((resolve) => {
-    diagnosisTimer = setTimeout(() => resolve({ checks: [{ name: 'Diagnosis', status: 'WARN', detail: `Timed out after ${GLOBAL_DIAGNOSIS_TIMEOUT_MS / 1000}s — some checks incomplete` }], issues: 0, timedOut: true }), GLOBAL_DIAGNOSIS_TIMEOUT_MS);
+    diagnosisTimer = setTimeout(() => {
+      const completed = Array.isArray(progress.checks) ? progress.checks : [];
+      const reported = new Set(completed.map((c) => c && c.name));
+      // Name the surfaces that did NOT report, so the reader knows exactly which
+      // conclusions are missing rather than just "something timed out".
+      const pending = ['Hooks', 'Adapters', 'Skills', 'Gate env', 'OpenCode TUI']
+        .filter((name) => !reported.has(name));
+      resolve(mergeTimedOutDiagnosis(
+        { checks: completed, issues: progress.getIssues() },
+        pending,
+      ));
+    }, GLOBAL_DIAGNOSIS_TIMEOUT_MS);
   });
   const { checks, issues: diagnosedIssues, timedOut } = await Promise.race([
     diagnosisPromise.then(r => ({ ...r, timedOut: false })),
@@ -3349,6 +3405,7 @@ module.exports = {
   readTuiJson,
   formatDoctorJson,
   diagnoseHookDrift,
+  mergeTimedOutDiagnosis,
   classifyHooksDir,
   syncGlobalHooksFromRepo,
   diagnoseModuleDrift,

package/lib/mattpocock-license-verdict.json

@@ -2,6 +2,6 @@
   "verdict": "PASS",
   "license_type": "MIT",
   "sha256": "0e7ac423bf2c6e223b7c5b156f8cf72da49d748e56a1641402c31f22ad07dbb5",
-  "checked_at": "2026-10-30T14:53:53.917Z",
+  "checked_at": "2026-10-30T15:31:30.899Z",
   "source": "network"
 }

package/mock-policy/AGENTS.md

@@ -1,9 +1,9 @@
 # SRC/MOCK-POLICY KNOWLEDGE BASE
 
 **Generated:** 2026-10-30
-**Commit:** e583bcfb
+**Commit:** c8a73b2a
 **Branch:** main
-**Version:** 0.34.1.0
+**Version:** 0.35.0.0
 
 ## OVERVIEW
 Mock layering policy enforcement — Gate M3 of pre-push hook. Ensures integration tests use real implementations for internal dependencies, mock external dependencies, and annotate pending mocks with removal plans. Combines project scope scanning, mock decision engine, and per-file validation into a single pipeline.

package/mutation/AGENTS.md

@@ -1,9 +1,9 @@
 # SRC/MUTATION KNOWLEDGE BASE
 
 **Generated:** 2026-10-30
-**Commit:** e583bcfb
+**Commit:** c8a73b2a
 **Branch:** main
-**Version:** 0.34.1.0
+**Version:** 0.35.0.0
 
 ## OVERVIEW
 **Gate M** (incremental mutation testing) + **Gate M2** helpers (test-layer detection used by `src/mock-policy/`). Pre-push quality gate. TypeScript-only; uses Stryker.

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@boyingliu01/xp-gate",
-  "version": "0.34.1",
+  "version": "0.35.0",
   "description": "AI-driven development workflow: 6 quality gates + Delphi review + Sprint Flow",
   "bin": {
     "xp-gate": "bin/xp-gate.js"

package/plugins/claude-code/.claude-plugin/plugin.json

@@ -1,6 +1,6 @@
 {
   "name": "xp-gate",
-  "version": "0.34.1",
+  "version": "0.35.0",
   "displayName": "XP-Gate",
   "description": "Extreme Programming quality gates + AI workflow skills for Claude Code. Includes 12 pre-commit gates (Gate 0-11), 8 pre-push gates, Sprint Flow (6 phases), and Delphi multi-expert review (>=90% consensus).",
   "author": {

package/plugins/claude-code/skills/delphi-review/AGENTS.md

@@ -1,9 +1,9 @@
 # SKILLS/DELPHI-REVIEW KNOWLEDGE BASE
 
 **Generated:** 2026-10-30
-**Commit:** e583bcfb
+**Commit:** c8a73b2a
 **Branch:** main
-**Version:** 0.34.1.0
+**Version:** 0.35.0.0
 
 ## OVERVIEW
 Delphi Consensus Review — multi-round anonymous expert review (≥90% threshold, exactly 3 experts with distinct executable model IDs). Supports design, requirements, and code-walkthrough modes.

package/plugins/claude-code/skills/sprint-flow/AGENTS.md

@@ -1,9 +1,9 @@
 # SKILLS/SPRINT-FLOW KNOWLEDGE BASE
 
 **Generated:** 2026-10-30
-**Commit:** e583bcfb
+**Commit:** c8a73b2a
 **Branch:** main
-**Version:** 0.34.1.0
+**Version:** 0.35.0.0
 
 ## OVERVIEW
 **6-phase** development pipeline (v2.0 compact redesign, Issue #290): PREP → DESIGN → BUILD → VERIFY → SHIP → CLOSE. Phase 3/6 BUILD default build mode is **ralph-loop** (REQ-level iteration, 40-67% token savings vs parallel). HARD-GATE between DESIGN (2/6) and BUILD (3/6): design must pass Delphi review (≥90% consensus) before any coding.

package/plugins/claude-code/skills/test-specification-alignment/AGENTS.md

@@ -1,9 +1,9 @@
 # SKILLS/TEST-SPECIFICATION-ALIGNMENT KNOWLEDGE BASE
 
 **Generated:** 2026-10-30
-**Commit:** e583bcfb
+**Commit:** c8a73b2a
 **Branch:** main
-**Version:** 0.34.1.0
+**Version:** 0.35.0.0
 
 ## OVERVIEW
 Test-Specification Alignment Engine — two-stage validation ensuring tests accurately reflect requirements and design specs.

package/plugins/dsh/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@boyingliu01/dsh-plugin-xp-gate",
-  "version": "0.34.1",
+  "version": "0.35.0",
   "description": "XP-Gate deterministic quality gates + AI workflow skills as a DeepSeek Harness (DSH) native plugin: gate-check / gate-principles / gate-arch tools + 12 SKILL.md skills",
   "type": "module",
   "main": "./lib/index.js",

package/plugins/dsh/skills/delphi-review/AGENTS.md

@@ -1,9 +1,9 @@
 # SKILLS/DELPHI-REVIEW KNOWLEDGE BASE
 
 **Generated:** 2026-10-30
-**Commit:** e583bcfb
+**Commit:** c8a73b2a
 **Branch:** main
-**Version:** 0.34.1.0
+**Version:** 0.35.0.0
 
 ## OVERVIEW
 Delphi Consensus Review — multi-round anonymous expert review (≥90% threshold, exactly 3 experts with distinct executable model IDs). Supports design, requirements, and code-walkthrough modes.

package/plugins/dsh/skills/sprint-flow/AGENTS.md

@@ -1,9 +1,9 @@
 # SKILLS/SPRINT-FLOW KNOWLEDGE BASE
 
 **Generated:** 2026-10-30
-**Commit:** e583bcfb
+**Commit:** c8a73b2a
 **Branch:** main
-**Version:** 0.34.1.0
+**Version:** 0.35.0.0
 
 ## OVERVIEW
 **6-phase** development pipeline (v2.0 compact redesign, Issue #290): PREP → DESIGN → BUILD → VERIFY → SHIP → CLOSE. Phase 3/6 BUILD default build mode is **ralph-loop** (REQ-level iteration, 40-67% token savings vs parallel). HARD-GATE between DESIGN (2/6) and BUILD (3/6): design must pass Delphi review (≥90% consensus) before any coding.

package/plugins/dsh/skills/test-specification-alignment/AGENTS.md

@@ -1,9 +1,9 @@
 # SKILLS/TEST-SPECIFICATION-ALIGNMENT KNOWLEDGE BASE
 
 **Generated:** 2026-10-30
-**Commit:** e583bcfb
+**Commit:** c8a73b2a
 **Branch:** main
-**Version:** 0.34.1.0
+**Version:** 0.35.0.0
 
 ## OVERVIEW
 Test-Specification Alignment Engine — two-stage validation ensuring tests accurately reflect requirements and design specs.

package/plugins/opencode/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@boyingliu01/opencode-plugin",
-  "version": "0.34.1",
+  "version": "0.35.0",
   "type": "module",
   "main": "index.ts",
   "description": "XP-Gate quality gates + AI workflow skills + Sprint Flow TUI sidebar for OpenCode",

package/plugins/opencode/skills/delphi-review/AGENTS.md

@@ -1,9 +1,9 @@
 # SKILLS/DELPHI-REVIEW KNOWLEDGE BASE
 
 **Generated:** 2026-10-30
-**Commit:** e583bcfb
+**Commit:** c8a73b2a
 **Branch:** main
-**Version:** 0.34.1.0
+**Version:** 0.35.0.0
 
 ## OVERVIEW
 Delphi Consensus Review — multi-round anonymous expert review (≥90% threshold, exactly 3 experts with distinct executable model IDs). Supports design, requirements, and code-walkthrough modes.

package/plugins/opencode/skills/sprint-flow/AGENTS.md

@@ -1,9 +1,9 @@
 # SKILLS/SPRINT-FLOW KNOWLEDGE BASE
 
 **Generated:** 2026-10-30
-**Commit:** e583bcfb
+**Commit:** c8a73b2a
 **Branch:** main
-**Version:** 0.34.1.0
+**Version:** 0.35.0.0
 
 ## OVERVIEW
 **6-phase** development pipeline (v2.0 compact redesign, Issue #290): PREP → DESIGN → BUILD → VERIFY → SHIP → CLOSE. Phase 3/6 BUILD default build mode is **ralph-loop** (REQ-level iteration, 40-67% token savings vs parallel). HARD-GATE between DESIGN (2/6) and BUILD (3/6): design must pass Delphi review (≥90% consensus) before any coding.

package/plugins/opencode/skills/test-specification-alignment/AGENTS.md

@@ -1,9 +1,9 @@
 # SKILLS/TEST-SPECIFICATION-ALIGNMENT KNOWLEDGE BASE
 
 **Generated:** 2026-10-30
-**Commit:** e583bcfb
+**Commit:** c8a73b2a
 **Branch:** main
-**Version:** 0.34.1.0
+**Version:** 0.35.0.0
 
 ## OVERVIEW
 Test-Specification Alignment Engine — two-stage validation ensuring tests accurately reflect requirements and design specs.

package/plugins/qoder/plugin.json

@@ -1,6 +1,6 @@
 {
   "name": "xp-gate",
-  "version": "0.34.1",
+  "version": "0.35.0",
   "displayName": "XP-Gate",
   "description": "Extreme Programming quality gates + AI workflow skills for Qoder. Includes 12 pre-commit gates (Gate 0-11), 8 pre-push gates, Sprint Flow (6 phases), and Delphi multi-expert review (>=90% consensus).",
   "author": {

package/plugins/qoder/skills/delphi-review/AGENTS.md

@@ -1,9 +1,9 @@
 # SKILLS/DELPHI-REVIEW KNOWLEDGE BASE
 
 **Generated:** 2026-10-30
-**Commit:** e583bcfb
+**Commit:** c8a73b2a
 **Branch:** main
-**Version:** 0.34.1.0
+**Version:** 0.35.0.0
 
 ## OVERVIEW
 Delphi Consensus Review — multi-round anonymous expert review (≥90% threshold, exactly 3 experts with distinct executable model IDs). Supports design, requirements, and code-walkthrough modes.

package/plugins/qoder/skills/sprint-flow/AGENTS.md

@@ -1,9 +1,9 @@
 # SKILLS/SPRINT-FLOW KNOWLEDGE BASE
 
 **Generated:** 2026-10-30
-**Commit:** e583bcfb
+**Commit:** c8a73b2a
 **Branch:** main
-**Version:** 0.34.1.0
+**Version:** 0.35.0.0
 
 ## OVERVIEW
 **6-phase** development pipeline (v2.0 compact redesign, Issue #290): PREP → DESIGN → BUILD → VERIFY → SHIP → CLOSE. Phase 3/6 BUILD default build mode is **ralph-loop** (REQ-level iteration, 40-67% token savings vs parallel). HARD-GATE between DESIGN (2/6) and BUILD (3/6): design must pass Delphi review (≥90% consensus) before any coding.

package/plugins/qoder/skills/test-specification-alignment/AGENTS.md

@@ -1,9 +1,9 @@
 # SKILLS/TEST-SPECIFICATION-ALIGNMENT KNOWLEDGE BASE
 
 **Generated:** 2026-10-30
-**Commit:** e583bcfb
+**Commit:** c8a73b2a
 **Branch:** main
-**Version:** 0.34.1.0
+**Version:** 0.35.0.0
 
 ## OVERVIEW
 Test-Specification Alignment Engine — two-stage validation ensuring tests accurately reflect requirements and design specs.

package/principles/AGENTS.md

@@ -1,9 +1,9 @@
 # PRINCIPLES CHECKER MODULE
 
 **Generated:** 2026-10-30
-**Commit:** e583bcfb
+**Commit:** c8a73b2a
 **Branch:** main
-**Version:** 0.34.1.0
+**Version:** 0.35.0.0
 
 ## OVERVIEW
 Clean Code & SOLID principles checker — **Gate 4** of pre-commit. 14 rules × 9 language adapters, SARIF 2.1.0 output. Houses the **Boy Scout Rule** enforcement engine (Gate 6) and warning-baseline storage.

package/skills/delphi-review/AGENTS.md

@@ -1,9 +1,9 @@
 # SKILLS/DELPHI-REVIEW KNOWLEDGE BASE
 
 **Generated:** 2026-10-30
-**Commit:** e583bcfb
+**Commit:** c8a73b2a
 **Branch:** main
-**Version:** 0.34.1.0
+**Version:** 0.35.0.0
 
 ## OVERVIEW
 Delphi Consensus Review — multi-round anonymous expert review (≥90% threshold, exactly 3 experts with distinct executable model IDs). Supports design, requirements, and code-walkthrough modes.

package/skills/sprint-flow/AGENTS.md

@@ -1,9 +1,9 @@
 # SKILLS/SPRINT-FLOW KNOWLEDGE BASE
 
 **Generated:** 2026-10-30
-**Commit:** e583bcfb
+**Commit:** c8a73b2a
 **Branch:** main
-**Version:** 0.34.1.0
+**Version:** 0.35.0.0
 
 ## OVERVIEW
 **6-phase** development pipeline (v2.0 compact redesign, Issue #290): PREP → DESIGN → BUILD → VERIFY → SHIP → CLOSE. Phase 3/6 BUILD default build mode is **ralph-loop** (REQ-level iteration, 40-67% token savings vs parallel). HARD-GATE between DESIGN (2/6) and BUILD (3/6): design must pass Delphi review (≥90% consensus) before any coding.

package/skills/test-specification-alignment/AGENTS.md

@@ -1,9 +1,9 @@
 # SKILLS/TEST-SPECIFICATION-ALIGNMENT KNOWLEDGE BASE
 
 **Generated:** 2026-10-30
-**Commit:** e583bcfb
+**Commit:** c8a73b2a
 **Branch:** main
-**Version:** 0.34.1.0
+**Version:** 0.35.0.0
 
 ## OVERVIEW
 Test-Specification Alignment Engine — two-stage validation ensuring tests accurately reflect requirements and design specs.