devlyn-cli 2.9.3 → 2.10.1

package/scripts/lint-skills.sh

@@ -60,17 +60,23 @@ _shared/spec-verify-check.py
 _shared/collect-codex-findings.py
 _shared/verify-merge-findings.py
 _shared/state-phase-write.py
+_shared/terminal-claim-check.py
+_shared/resolve-stop-hook.py
+_shared/resolve-bootstrap.py
 devlyn:ideate/SKILL.md
 devlyn:ideate/references/spec-template.md
 devlyn:ideate/references/elicitation.md
 devlyn:ideate/references/project-mode.md
 devlyn:ideate/references/from-spec-mode.md
 devlyn:resolve/SKILL.md
+devlyn:queue/SKILL.md
+devlyn:engines/SKILL.md
 devlyn:resolve/references/state-schema.md
 devlyn:resolve/references/free-form-mode.md
 devlyn:resolve/references/phases/plan.md
 devlyn:resolve/references/phases/probe-derive.md
 devlyn:resolve/references/phases/implement.md
+devlyn:resolve/references/phases/surface-close.md
 devlyn:resolve/references/phases/build-gate.md
 devlyn:resolve/references/phases/cleanup.md
 devlyn:resolve/references/phases/verify.md
@@ -78,6 +84,7 @@ _shared/expected.schema.json
 _shared/adapters/README.md
 _shared/adapters/claude.md
 _shared/adapters/codex.md
+_shared/adapters/grok.md
 _shared/adapters/omp.md
 _shared/codex-config.md
 _shared/codex-monitored.sh
@@ -146,6 +153,7 @@ offenders=$(git grep -Il -- 'mcp__codex-cli__' -- \
   ':!config/skills/preflight-workspace/**' \
   ':!benchmark/auto-resolve/external/**' \
   ':!benchmark/auto-resolve/results/**' \
+  ':!benchmark/ceiling/results/**' \
   ':!benchmark/auto-resolve/PILOT-RESULTS*' \
   2>/dev/null || true)
 if [ -z "$offenders" ]; then
@@ -170,6 +178,7 @@ offenders=$(git grep -Il -- 'Requires Codex MCP\|Codex MCP server\|Codex MCP ava
   ':!config/skills/preflight-workspace/**' \
   ':!benchmark/auto-resolve/external/**' \
   ':!benchmark/auto-resolve/results/**' \
+  ':!benchmark/ceiling/results/**' \
   ':!benchmark/auto-resolve/PILOT-RESULTS*' \
   2>/dev/null || true)
 if [ -z "$offenders" ]; then
@@ -198,16 +207,19 @@ else
 fi
 
 # ---------------------------------------------------------------------------
-# 4. No model-pinned Claude references, any generation. Engine-scope docs say
-#    "Claude"; exact model ids belong in run metadata only.
+# 4. No model-pinned Claude references, any generation, except the adjudicated
+#    SURFACE_CLOSE envelope whose requested id must equal CLI arg and modelUsage.
 # ---------------------------------------------------------------------------
 section "Check 4: No model-pinned Claude references"
-offenders=$(grep -RIlnE 'Claude (Opus|Sonnet|Haiku|Fable)|claude-(opus|sonnet|haiku|fable)-[0-9]' \
+offenders=$(grep -RInE 'Claude (Opus|Sonnet|Haiku|Fable)|claude-(opus|sonnet|haiku|fable)-[0-9]' \
   config/skills 2>/dev/null \
   | grep -v 'config/skills/roadmap-archival-workspace/' \
   | grep -v 'config/skills/devlyn:auto-resolve-workspace/' \
   | grep -v 'config/skills/devlyn:ideate-workspace/' \
   | grep -v 'config/skills/preflight-workspace/' \
+  | grep -vE '^config/skills/devlyn:resolve/SKILL\.md:[0-9]+:Freeze .*--tools "Read,Grep,Glob,Edit,Write" --dangerously-skip-permissions --model claude-sonnet-5 --output-format json --strict-mcp-config --mcp-config '\''\{"mcpServers":\{\}\}'\''.*, recording the same model at SPW spawn\.' \
+  | cut -d: -f1 \
+  | sort -u \
   || true)
 if [ -z "$offenders" ]; then
   ok "no model-pinned Claude references"
@@ -257,6 +269,84 @@ else
   bad "devlyn:design-ui must not be listed in OPTIONAL_ADDONS"
 fi
 
+# ---------------------------------------------------------------------------
+# 5b. Every completed managed-skills install writes its own version marker.
+# ---------------------------------------------------------------------------
+section "Check 5b: Installed skill roots carry exact version markers"
+if make_temp_dir tmp_install_marker /tmp/devlyn-install-marker.XXXXXX; then
+  marker_home="$tmp_install_marker/home"
+  marker_project="$tmp_install_marker/project"
+  installer="$PWD/bin/devlyn.js"
+  mkdir -p "$marker_home" "$marker_project"
+
+  if (cd "$marker_project" \
+      && HOME="$marker_home" node "$installer" -y >"$tmp_install_marker/claude.log" 2>&1 \
+      && HOME="$marker_home" node "$installer" agents all >"$tmp_install_marker/agents.log" 2>&1); then
+    if python3 - "$installer" "$marker_home" "$marker_project" <<'PY'
+import json
+import pathlib
+import stat
+import sys
+
+installer = pathlib.Path(sys.argv[1])
+home = pathlib.Path(sys.argv[2])
+project = pathlib.Path(sys.argv[3])
+package = json.loads((installer.parent.parent / "package.json").read_text())
+expected = {"schemaVersion": 1, "package": package["name"], "version": package["version"]}
+roots = [
+    project / ".claude" / "skills",
+    home / ".codex" / "skills",
+    home / ".agents" / "skills",
+    home / ".grok" / "skills",
+]
+for root in roots:
+    marker = root / ".devlyn-install.json"
+    actual = json.loads(marker.read_text())
+    if actual != expected:
+        raise SystemExit(f"{marker}: expected {expected}, got {actual}")
+    if stat.S_IMODE(marker.stat().st_mode) != 0o600:
+        raise SystemExit(f"{marker}: expected mode 0600")
+    if list(root.glob(".devlyn-install.json.*.tmp")):
+        raise SystemExit(f"{root}: stale marker temp file")
+PY
+    then
+      if grep -Fxq '.claude/skills/.devlyn-install.json' "$marker_project/.gitignore"; then
+        ok "Claude, Codex, shared-agent, and Grok roots have exact 0600 markers; project marker is ignored"
+      else
+        bad "Claude project install must ignore its local marker"
+      fi
+    else
+      bad "completed installs must write the exact per-root marker"
+    fi
+  else
+    bad "isolated managed-skills install failed"
+  fi
+
+  incomplete="$tmp_install_marker/incomplete"
+  mkdir -p "$incomplete/package" "$incomplete/home/.codex/skills" "$incomplete/claude/.claude/skills" "$incomplete/codex"
+  cp -R bin config package.json "$incomplete/package/"
+  rm -rf "$incomplete/package/config/skills/devlyn:queue"
+  printf '{"version":"stale"}\n' > "$incomplete/home/.codex/skills/.devlyn-install.json"
+  printf '{"version":"stale"}\n' > "$incomplete/claude/.claude/skills/.devlyn-install.json"
+  if ! (cd "$incomplete/claude" \
+      && HOME="$incomplete/home" node "$incomplete/package/bin/devlyn.js" -y >"$incomplete/claude.log" 2>&1) \
+      && ! (cd "$incomplete/codex" \
+      && HOME="$incomplete/home" node "$incomplete/package/bin/devlyn.js" agents codex >"$incomplete/codex.log" 2>&1) \
+      && grep -Fq 'Incomplete devlyn skill install; missing: devlyn:queue' "$incomplete/claude.log" \
+      && grep -Fq 'Incomplete devlyn skill install; missing: devlyn:queue' "$incomplete/codex.log" \
+      && [ -d "$incomplete/claude/.claude/skills/devlyn:resolve" ] \
+      && [ -d "$incomplete/home/.codex/skills/devlyn:resolve" ] \
+      && [ ! -e "$incomplete/claude/.claude/skills/.devlyn-install.json" ] \
+      && [ ! -e "$incomplete/home/.codex/skills/.devlyn-install.json" ]; then
+    ok "incomplete Claude and Codex copies fail visibly without stale or replacement markers"
+  else
+    bad "incomplete Claude and Codex copies must fail visibly without an install marker"
+  fi
+  rm -rf "$tmp_install_marker"
+else
+  bad "could not allocate install-marker smoke directory"
+fi
+
 # ---------------------------------------------------------------------------
 # 6. Source ↔ installed mirror parity on critical path.
 # Only runs if .claude/skills exists (i.e. installer has been run).
@@ -332,23 +422,23 @@ if ! grep -Fq 'def pair_trigger_skip_contract_violation' config/skills/_shared/v
   || ! grep -Fq 'verify-pair-trigger-skipped-reason-unsupported' config/skills/_shared/verify-merge-findings.py \
   || ! grep -Fq 'verify-pair-trigger-mechanical-blocker-unsupported' config/skills/_shared/verify-merge-findings.py \
   || ! grep -Fq 'verify-pair-trigger-primary-judge-blocker-unsupported' config/skills/_shared/verify-merge-findings.py \
-  || ! grep -Fq 'eligible:false` with no supported skip reason' config/skills/devlyn:resolve/references/state-schema.md \
-  || ! grep -Fq 'Canonical eligible reasons are `mode.verify-only`' config/skills/devlyn:resolve/references/state-schema.md \
+  || ! grep -Fq 'Missing, contradictory, incomplete, or unknown trigger state BLOCKs VERIFY' config/skills/devlyn:resolve/references/state-schema.md \
+  || ! grep -Fq 'Canonical reasons are `pair.default`, `mode.verify-only`' config/skills/devlyn:resolve/references/state-schema.md \
   || ! grep -Fq '`mode.pair-verify`' config/skills/devlyn:resolve/references/state-schema.md \
-  || ! grep -Fq 'Eligible triggers must contain only canonical' config/skills/devlyn:resolve/references/phases/verify.md \
-  || ! grep -Fq 'include every applicable canonical reason' config/skills/devlyn:resolve/references/phases/verify.md \
+  || ! grep -Fq 'Eligible schema-v3 state must contain `pair.default` and every other applicable canonical' config/skills/devlyn:resolve/references/phases/verify.md \
+  || ! grep -Fq 'with `pair.default` plus every applicable outcome-independent canonical reason' config/skills/devlyn:resolve/references/phases/verify.md \
   || ! grep -Fq '`mode.pair-verify`' config/skills/devlyn:resolve/references/phases/verify.md \
   || ! grep -Fq '"pair_verify": false' config/skills/devlyn:resolve/references/state-schema.md \
-  || ! grep -Fq 'requires a non-empty reasons list containing every applicable canonical eligible reason' config/skills/devlyn:resolve/references/state-schema.md \
-  || ! grep -Fq 'containing every applicable canonical eligible reason' config/skills/devlyn:resolve/references/state-schema.md \
-  || ! grep -Fq 'state.pair_verify == true' config/skills/devlyn:resolve/references/state-schema.md \
+  || ! grep -Fq 'Schema v3.0 eligible state requires `pair.default` plus every applicable telemetry reason' config/skills/devlyn:resolve/references/state-schema.md \
+  || ! grep -Fq 'Ineligible new-run state has empty reasons and only' config/skills/devlyn:resolve/references/state-schema.md \
+  || ! grep -Fq 'True only for `--pair-verify`' config/skills/devlyn:resolve/references/state-schema.md \
   || ! grep -Fq 'pair_verify: true` only when `--pair-verify` was passed' config/skills/devlyn:resolve/SKILL.md \
-  || ! grep -Fq 'include every applicable canonical reason' config/skills/devlyn:resolve/SKILL.md \
+  || ! grep -Fq 'persist `pair_trigger` before spawn with `pair.default` plus every applicable outcome-independent reason' config/skills/devlyn:resolve/SKILL.md \
   || ! grep -Fq '`--pair-verify` and `--no-pair` are mutually exclusive' config/skills/devlyn:resolve/SKILL.md \
   || ! grep -Fq 'mutually exclusive with `risk_profile.pair_default_enabled == false`' config/skills/devlyn:resolve/references/state-schema.md \
   || ! grep -Fq 'if both are present, stop with `BLOCKED:invalid-flags`' config/skills/devlyn:resolve/references/phases/verify.md \
-  || ! grep -Fq 'Contradictory, incomplete, or unknown trigger state is a VERIFY contract violation' config/skills/devlyn:resolve/SKILL.md \
-  || ! grep -Fq 'user_no_pair` is valid only when `risk_profile.pair_default_enabled == false`' config/skills/devlyn:resolve/references/state-schema.md \
+  || ! grep -Fq 'Malformed risk or trigger state BLOCKs VERIFY' config/skills/devlyn:resolve/SKILL.md \
+  || ! grep -Fq 'pair_default_enabled` is false only for explicit `--no-pair`' config/skills/devlyn:resolve/references/state-schema.md \
   || ! grep -Fq 'def reject_json_constant' config/skills/_shared/verify-merge-findings.py \
   || ! grep -Fq 'loads_strict_json(raw)' config/skills/_shared/verify-merge-findings.py \
   || ! grep -Fq 'invalid JSON numeric constant: NaN' config/skills/_shared/verify-merge-findings.py; then
@@ -377,9 +467,17 @@ if python3 config/skills/_shared/collect-codex-findings.py --self-test >/dev/nul
 else
   bad "collect-codex-findings.py self-test failed"
 fi
+# iter-0082: the frozen collector contract — every negative through every ingress
+# path, plus real judge captures replayed against hand-verified expectations.
+# The self-test above is unit-shaped; this is the integration bar against real bytes.
+if python3 benchmark/ceiling/probes/r-weld-0082/test-collector-contract.py >/dev/null 2>&1; then
+  ok "collector contract (iter-0082 negatives + real-capture corpus) passed"
+else
+  bad "collector contract failed — run benchmark/ceiling/probes/r-weld-0082/test-collector-contract.py"
+fi
 if ! grep -Fq 'def reject_json_constant' config/skills/_shared/collect-codex-findings.py \
   || ! grep -Fq 'loads_strict_json(raw)' config/skills/_shared/collect-codex-findings.py \
-  || ! grep -Fq 'NaN Codex stdout finding must not normalize' config/skills/_shared/collect-codex-findings.py; then
+  || ! grep -Fq 'NaN pair-JUDGE stdout finding' config/skills/_shared/collect-codex-findings.py; then
   bad "collect-codex-findings.py must reject non-standard JSON constants in pair-JUDGE stdout"
 fi
 
@@ -396,10 +494,56 @@ if ! grep -Fq 'SAFE_RUN_ID_RE' config/skills/_shared/archive_run.py \
   || ! grep -Fq 'invalid JSON numeric constant: NaN' config/skills/_shared/archive_run.py \
   || ! grep -Fq '"verify.pair.findings.jsonl"' config/skills/_shared/archive_run.py \
   || ! grep -Fq '"verify-merge.summary.json"' config/skills/_shared/archive_run.py \
+  || ! grep -Fq '"surface-close.output.json"' config/skills/_shared/archive_run.py \
   || ! grep -Fq '"*-judge.*"' config/skills/_shared/archive_run.py; then
   bad "archive_run.py must safely archive pair/risk-probe evidence and reject unsafe run ids"
 fi
 
+section "Check 6c2: SURFACE_CLOSE v6 dispatch is complete and mirrored"
+surface_close_missing=0
+if python3 config/skills/_shared/state-phase-write.py --self-test >/dev/null 2>&1; then
+  ok "SURFACE_CLOSE state/envelope/adjudication/audit/rollback self-test passed"
+else
+  bad "state-phase-write.py SURFACE_CLOSE self-test failed"
+fi
+if ! grep -Fq 'rollback_surface_delta' config/skills/_shared/state-phase-write.py \
+  || ! grep -Fq 'validate_surface_adjudication' config/skills/_shared/state-phase-write.py \
+  || ! grep -Fq 'validate_surface_execution' config/skills/_shared/state-phase-write.py \
+  || ! grep -Fq 'phases.surface_close spawn requires --engine claude' config/skills/_shared/state-phase-write.py \
+  || ! grep -Fq 'phases.surface_close spawn requires --model' config/skills/_shared/state-phase-write.py \
+  || ! grep -Fq 'BLOCKED:surface-close-input-mismatch' config/skills/_shared/state-phase-write.py \
+  || ! grep -Fq 'surface-check' config/skills/_shared/state-phase-write.py \
+  || ! grep -Fq 'surface-rollback' config/skills/_shared/state-phase-write.py; then
+  bad "state-phase-write.py must mechanically adjudicate, audit, guard, and roll back SURFACE_CLOSE"
+fi
+for tree in config/skills .claude/skills .agents/skills; do
+  skill="$tree/devlyn:resolve/SKILL.md"
+  phase="$tree/devlyn:resolve/references/phases/surface-close.md"
+  if ! grep -Fq '## PHASE 2.5: SURFACE_CLOSE' "$skill" \
+    || ! grep -Fq '`state.source.type == "generated"` and complexity is trivial/medium' "$skill" \
+    || ! grep -Fq 'Engine is Claude always' "$skill" \
+    || ! grep -Fq 'executor flag/pin' "$skill" \
+    || ! grep -Fq 'auto_surface_close_claude_unavailable' "$skill" \
+    || ! grep -Fq 'canonical body VERBATIM' "$skill" \
+    || ! grep -Fq 'run-bounded.py 600 -- claude -p' "$skill" \
+    || ! grep -Fq -- '--tools "Read,Grep,Glob,Edit,Write" --dangerously-skip-permissions --model claude-sonnet-5 --output-format json --strict-mcp-config --mcp-config '\''{"mcpServers":{}}'\''' "$skill" \
+    || ! grep -Fq '.devlyn/surface-close.output.json' "$skill" \
+    || ! grep -Fq '**Common post-fix checkpoint (BUILD_GATE and VERIFY):**' "$skill" \
+    || ! grep -Fq 'durability-enforce --round <n> --origin-phase <build_gate|verify>' "$skill" \
+    || ! grep -Fq '`surface-rollback`' "$skill" \
+    || grep -Fq 'Supplied digests' "$phase" \
+    || grep -Fq 'Hash both artifacts first' "$phase" \
+    || ! grep -Fq 'Never modify inputs or read state, PLAN, or IMPLEMENT transcript/reasoning.' "$phase" \
+    || ! grep -Fq 'optionally followed by ` — <one-line evidence>`' "$phase" \
+    || ! grep -Fq 'N/A <authorized-file>[:<line>] — <one-line evidence-based relationship judgment>' "$phase"; then
+    bad "$tree — SURFACE_CLOSE v6 dispatch/timeout/adjudication/rollback contract missing"
+    surface_close_missing=1
+  fi
+done
+if [ $surface_close_missing -eq 0 ]; then
+  ok "SURFACE_CLOSE v6 dispatch/timeout/adjudication/rollback contract is mirrored"
+fi
+
 section "Check 6d: Spec verification executes hidden-blind risk probes"
 if python3 config/skills/_shared/spec-verify-check.py --self-test >/dev/null 2>&1; then
   ok "spec-verify-check.py risk-probe and expected-contract self-test passed"
@@ -647,13 +791,16 @@ fi
 if ! grep -Fq '"uses_visible_input_key_names"' config/skills/_shared/spec-verify-check.py \
   || ! grep -Fq '"asserts_visible_output_key_names"' config/skills/_shared/spec-verify-check.py \
   || ! grep -Fq '"asserts_no_unexpected_output_keys"' config/skills/_shared/spec-verify-check.py \
-  || ! grep -Fq 'shape_contract co-occurring with error_contract without' config/skills/_shared/spec-verify-check.py \
+  || ! grep -Fq 'shape_contract claiming visible exact error object without' config/skills/_shared/spec-verify-check.py \
+  || ! grep -Fq 'visible_text_names_exact_json_error_object' config/skills/_shared/spec-verify-check.py \
   || ! grep -Fq 'JSON error object shape_contract with exact object evidence was rejected' config/skills/_shared/spec-verify-check.py \
   || ! grep -Fq 'shape_contract without any evidence was accepted' config/skills/_shared/spec-verify-check.py \
   || ! grep -Fq 'shape_contract with exact key evidence was rejected' config/skills/_shared/spec-verify-check.py \
   || ! grep -Fq '`shape_contract` when the visible text names exact keys' config/skills/devlyn:resolve/references/phases/probe-derive.md \
-  || ! grep -Fq '`shape_contract` must' config/skills/devlyn:resolve/SKILL.md; then
-  bad "risk-probe shape contracts must require exact visible input/output key evidence unconditionally, and asserts_exact_error_object when co-occurring with error_contract/http_error_contract"
+  || ! grep -Fq 'visible_text_names_exact_json_error_object' config/skills/devlyn:resolve/references/phases/probe-derive.md \
+  || ! grep -Fq '`shape_contract` must' config/skills/devlyn:resolve/SKILL.md \
+  || ! grep -Fq 'visible_text_names_exact_json_error_object' config/skills/devlyn:resolve/SKILL.md; then
+  bad "risk-probe shape contracts must require exact visible input/output key evidence unconditionally, and asserts_exact_error_object when the probe claims the visible text names an exact JSON error object"
 fi
 if ! grep -Fq '"auth_signature_contract"' config/skills/_shared/spec-verify-check.py \
   || ! grep -Fq '"idempotency_replay"' config/skills/_shared/spec-verify-check.py \
@@ -960,7 +1107,7 @@ fi
 if ! grep -Fq 'per-feature Verification must include a solo-headroom' config/skills/devlyn:ideate/references/project-mode.md \
   || ! grep -Fq 'per-feature Verification must include a solo-headroom' .claude/skills/devlyn:ideate/references/project-mode.md \
   || ! grep -Fq 'per-feature Verification must include a solo-headroom' .agents/skills/devlyn:ideate/references/project-mode.md \
-  || rg -q 'Context or Verification' config/skills/devlyn:ideate/references .claude/skills/devlyn:ideate/references .agents/skills/devlyn:ideate/references; then
+  || grep -rq 'Context or Verification' config/skills/devlyn:ideate/references .claude/skills/devlyn:ideate/references .agents/skills/devlyn:ideate/references; then
   bad "ideate project mode must keep solo-headroom hypothesis inside each pair-candidate feature spec"
 else
   ok "ideate project mode keeps solo-headroom hypothesis inside each pair-candidate feature spec"
@@ -1078,38 +1225,39 @@ if ! grep -Fq 'If the spec includes a solo-headroom hypothesis, one of the two t
 else
   ok "resolve pair-JUDGE prompts prioritize solo-headroom hypothesis"
 fi
-if ! grep -Fq 'names the primary judge; if pair-mode triggers, the first available OTHER' config/skills/devlyn:resolve/references/phases/verify.md \
-  || ! grep -Fq 'names the primary judge; if pair-mode triggers, the first available OTHER' .claude/skills/devlyn:resolve/references/phases/verify.md \
-  || ! grep -Fq 'names the primary judge; if pair-mode triggers, the first available OTHER' .agents/skills/devlyn:resolve/references/phases/verify.md \
-  || ! grep -Fq 'do not run the OTHER engine as the' config/skills/devlyn:resolve/SKILL.md \
-  || ! grep -Fq 'do not run the OTHER engine as the' .claude/skills/devlyn:resolve/SKILL.md \
-  || ! grep -Fq 'do not run the OTHER engine as the' .agents/skills/devlyn:resolve/SKILL.md \
-  || ! grep -Fq '_shared/adapters/claude.md` `## Invocation' config/skills/devlyn:resolve/references/phases/verify.md \
-  || ! grep -Fq '_shared/adapters/claude.md` `## Invocation' config/skills/devlyn:resolve/SKILL.md \
+if ! grep -Fq 'The `--engine` flag does not disable default pairing' config/skills/devlyn:resolve/references/phases/verify.md \
+  || ! grep -Fq 'The `--engine` flag does not disable default pairing' .claude/skills/devlyn:resolve/references/phases/verify.md \
+  || ! grep -Fq 'The `--engine` flag does not disable default pairing' .agents/skills/devlyn:resolve/references/phases/verify.md \
+  || ! grep -Fq 'the second judge uses the OTHER engine by default when available' config/skills/devlyn:resolve/SKILL.md \
+  || ! grep -Fq 'the second judge uses the OTHER engine by default when available' .claude/skills/devlyn:resolve/SKILL.md \
+  || ! grep -Fq 'the second judge uses the OTHER engine by default when available' .agents/skills/devlyn:resolve/SKILL.md \
+  || ! grep -Fq '_shared/adapters/<name>.md' config/skills/devlyn:resolve/references/phases/verify.md \
+  || ! grep -Fq '`## Invocation`' config/skills/devlyn:resolve/references/phases/verify.md \
+  || ! grep -Fq '_shared/adapters/<name>.md` `## Invocation' config/skills/devlyn:resolve/SKILL.md \
   || ! grep -Fq '## Invocation' config/skills/_shared/adapters/claude.md \
   || ! grep -Fq '.devlyn/claude-judge.stdout' config/skills/_shared/adapters/claude.md; then
   bad "engine-neutral pair-judge dual declaration (iter-0060) out of sync"
 else
   ok "engine-neutral pair-judge dual declaration (iter-0060) in sync"
 fi
-if ! grep -Fq 'current free-form `state.complexity` is `"large"`' config/skills/devlyn:resolve/SKILL.md \
-  || ! grep -Fq 'current free-form `state.complexity` is `"large"`' .claude/skills/devlyn:resolve/SKILL.md \
-  || ! grep -Fq 'current free-form `state.complexity` is `"large"`' .agents/skills/devlyn:resolve/SKILL.md \
-  || ! grep -Fq 'legacy/external spec `complexity: large` is accepted for compatibility' config/skills/devlyn:resolve/SKILL.md \
-  || ! grep -Fq 'legacy/external spec `complexity: large` is accepted for compatibility' .claude/skills/devlyn:resolve/SKILL.md \
-  || ! grep -Fq 'legacy/external spec `complexity: large` is accepted for compatibility' .agents/skills/devlyn:resolve/SKILL.md \
-  || ! grep -Fq 'current free-form `state.complexity` of `"large"`' config/skills/devlyn:resolve/references/state-schema.md \
-  || ! grep -Fq 'current free-form `state.complexity` of `"large"`' .claude/skills/devlyn:resolve/references/state-schema.md \
-  || ! grep -Fq 'current free-form `state.complexity` of `"large"`' .agents/skills/devlyn:resolve/references/state-schema.md \
-  || ! grep -Fq 'Legacy/external spec `complexity: large` remains accepted for compatibility' config/skills/devlyn:resolve/references/state-schema.md \
-  || ! grep -Fq 'Legacy/external spec `complexity: large` remains accepted for compatibility' .claude/skills/devlyn:resolve/references/state-schema.md \
-  || ! grep -Fq 'Legacy/external spec `complexity: large` remains accepted for compatibility' .agents/skills/devlyn:resolve/references/state-schema.md \
-  || ! grep -Fq 'Current free-form `state.complexity` is `"large"`' config/skills/devlyn:resolve/references/phases/verify.md \
-  || ! grep -Fq 'Current free-form `state.complexity` is `"large"`' .claude/skills/devlyn:resolve/references/phases/verify.md \
-  || ! grep -Fq 'Current free-form `state.complexity` is `"large"`' .agents/skills/devlyn:resolve/references/phases/verify.md \
-  || ! grep -Fq 'legacy/external spec' config/skills/devlyn:resolve/references/phases/verify.md \
-  || ! grep -Fq 'legacy/external spec' .claude/skills/devlyn:resolve/references/phases/verify.md \
-  || ! grep -Fq 'legacy/external spec' .agents/skills/devlyn:resolve/references/phases/verify.md; then
+if ! grep -Fq '`complexity.high`, `complexity.large`' config/skills/devlyn:resolve/SKILL.md \
+  || ! grep -Fq '`complexity.high`, `complexity.large`' .claude/skills/devlyn:resolve/SKILL.md \
+  || ! grep -Fq '`complexity.high`, `complexity.large`' .agents/skills/devlyn:resolve/SKILL.md \
+  || ! grep -Fq '`spec.complexity.high`, `spec.complexity.large`' config/skills/devlyn:resolve/SKILL.md \
+  || ! grep -Fq '`spec.complexity.high`, `spec.complexity.large`' .claude/skills/devlyn:resolve/SKILL.md \
+  || ! grep -Fq '`spec.complexity.high`, `spec.complexity.large`' .agents/skills/devlyn:resolve/SKILL.md \
+  || ! grep -Fq 'Legacy complexity values remain accepted only for archived compatibility' config/skills/devlyn:resolve/references/state-schema.md \
+  || ! grep -Fq 'Legacy complexity values remain accepted only for archived compatibility' .claude/skills/devlyn:resolve/references/state-schema.md \
+  || ! grep -Fq 'Legacy complexity values remain accepted only for archived compatibility' .agents/skills/devlyn:resolve/references/state-schema.md \
+  || ! grep -Fq '`complexity.high`, `complexity.large`, `spec.complexity.high`, `spec.complexity.large`' config/skills/devlyn:resolve/references/state-schema.md \
+  || ! grep -Fq '`complexity.high`, `complexity.large`, `spec.complexity.high`, `spec.complexity.large`' .claude/skills/devlyn:resolve/references/state-schema.md \
+  || ! grep -Fq '`complexity.high`, `complexity.large`, `spec.complexity.high`, `spec.complexity.large`' .agents/skills/devlyn:resolve/references/state-schema.md \
+  || ! grep -Fq '`complexity.large`, `spec.complexity.high`, `spec.complexity.large`,' config/skills/devlyn:resolve/references/phases/verify.md \
+  || ! grep -Fq '`complexity.large`, `spec.complexity.high`, `spec.complexity.large`,' .claude/skills/devlyn:resolve/references/phases/verify.md \
+  || ! grep -Fq '`complexity.large`, `spec.complexity.high`, `spec.complexity.large`,' .agents/skills/devlyn:resolve/references/phases/verify.md \
+  || ! grep -Fq '`spec.solo_headroom_hypothesis`, `risk.high`, `risk_probes.enabled`,' config/skills/devlyn:resolve/references/phases/verify.md \
+  || ! grep -Fq '`spec.solo_headroom_hypothesis`, `risk.high`, `risk_probes.enabled`,' .claude/skills/devlyn:resolve/references/phases/verify.md \
+  || ! grep -Fq '`spec.solo_headroom_hypothesis`, `risk.high`, `risk_probes.enabled`,' .agents/skills/devlyn:resolve/references/phases/verify.md; then
   bad "resolve VERIFY docs must distinguish current large complexity, legacy high state, and legacy large spec compatibility"
 else
   ok "resolve VERIFY docs distinguish current large complexity, legacy high state, and legacy large spec compatibility"
@@ -1165,15 +1313,15 @@ if ! grep -Fq 'def spec_has_solo_headroom_hypothesis' config/skills/_shared/veri
   || ! grep -Fq 'spec_has_solo_headroom_hypothesis(' .agents/skills/_shared/verify-merge-findings.py \
   || ! grep -Fq ') is False' .agents/skills/_shared/verify-merge-findings.py \
   || ! grep -Fq ') is True' .agents/skills/_shared/verify-merge-findings.py \
-  || ! grep -Fq 'The spec includes an actionable solo-headroom hypothesis.' \
+  || ! grep -Fq '`spec.solo_headroom_hypothesis`' \
     config/skills/devlyn:resolve/references/phases/verify.md \
-  || ! grep -Fq 'The spec includes an actionable solo-headroom hypothesis.' \
+  || ! grep -Fq '`spec.solo_headroom_hypothesis`' \
     .claude/skills/devlyn:resolve/references/phases/verify.md \
-  || ! grep -Fq 'The spec includes an actionable solo-headroom hypothesis.' \
+  || ! grep -Fq '`spec.solo_headroom_hypothesis`' \
     .agents/skills/devlyn:resolve/references/phases/verify.md \
-  || ! grep -Fq 'actionable solo-headroom hypotheses' config/skills/devlyn:resolve/references/state-schema.md \
-  || ! grep -Fq 'actionable solo-headroom hypotheses' .claude/skills/devlyn:resolve/references/state-schema.md \
-  || ! grep -Fq 'actionable solo-headroom hypotheses' .agents/skills/devlyn:resolve/references/state-schema.md \
+  || ! grep -Fq '`spec.solo_headroom_hypothesis`' config/skills/devlyn:resolve/references/state-schema.md \
+  || ! grep -Fq '`spec.solo_headroom_hypothesis`' .claude/skills/devlyn:resolve/references/state-schema.md \
+  || ! grep -Fq '`spec.solo_headroom_hypothesis`' .agents/skills/devlyn:resolve/references/state-schema.md \
   || ! grep -Fq 'same actionable solo-headroom hypothesis is a VERIFY pair-trigger reason' config/skills/devlyn:resolve/SKILL.md \
   || ! grep -Fq 'same actionable solo-headroom hypothesis is a VERIFY pair-trigger reason' .claude/skills/devlyn:resolve/SKILL.md \
   || ! grep -Fq 'same actionable solo-headroom hypothesis is a VERIFY pair-trigger reason' .agents/skills/devlyn:resolve/SKILL.md; then
@@ -1199,9 +1347,9 @@ if ! grep -Fq 'pair_evidence_intent' config/skills/devlyn:resolve/references/fre
   || ! grep -Fq 'state.source.criteria_sha256` for generated free-form mode' config/skills/devlyn:resolve/references/phases/verify.md \
   || ! grep -Fq 'state.source.criteria_sha256` for generated free-form mode' .claude/skills/devlyn:resolve/references/phases/verify.md \
   || ! grep -Fq 'state.source.criteria_sha256` for generated free-form mode' .agents/skills/devlyn:resolve/references/phases/verify.md \
-  || ! grep -Fq 'Free-form mode sets `type: "generated"`' config/skills/devlyn:resolve/references/state-schema.md \
-  || ! grep -Fq 'Free-form mode sets `type: "generated"`' .claude/skills/devlyn:resolve/references/state-schema.md \
-  || ! grep -Fq 'Free-form mode sets `type: "generated"`' .agents/skills/devlyn:resolve/references/state-schema.md \
+  || ! grep -Fq 'Free-form sets `type: "generated"`' config/skills/devlyn:resolve/references/state-schema.md \
+  || ! grep -Fq 'Free-form sets `type: "generated"`' .claude/skills/devlyn:resolve/references/state-schema.md \
+  || ! grep -Fq 'Free-form sets `type: "generated"`' .agents/skills/devlyn:resolve/references/state-schema.md \
   || ! grep -Fq 'backticked observable command line that itself contains `miss`' config/skills/devlyn:resolve/references/free-form-mode.md \
   || ! grep -Fq 'backticked observable command line that itself contains `miss`' .claude/skills/devlyn:resolve/references/free-form-mode.md \
   || ! grep -Fq 'backticked observable command line that itself contains `miss`' .agents/skills/devlyn:resolve/references/free-form-mode.md \
@@ -1311,19 +1459,21 @@ if [ $verify_mech_missing -eq 0 ]; then
   ok "spec-verify VERIFY output routes into verify-merge-findings.py"
 fi
 
-section "Check 6j: VERIFY pair trigger runs after primary JUDGE"
+section "Check 6j: VERIFY default pair dispatches concurrently"
 pair_trigger_order_missing=0
 for file in \
   config/skills/devlyn:resolve/SKILL.md \
   .claude/skills/devlyn:resolve/SKILL.md \
   .agents/skills/devlyn:resolve/SKILL.md
 do
-  if ! grep -Fq 'Pair-mode (cross-model JUDGE) is eligible only after MECHANICAL and the primary JUDGE have no verdict-binding findings' "$file" \
-    || ! grep -Fq 'After MECHANICAL and the primary JUDGE finish, compute `pair_trigger' "$file" \
-    || ! grep -Fq '`risk_profile` is strict typed state' "$file" \
-    || ! grep -Fq 'malformed `risk_profile` is also a VERIFY contract violation' "$file" \
-    || ! grep -Fq 'If MECHANICAL or the primary JUDGE has a verdict-binding finding' "$file"; then
-    bad "$file — VERIFY pair trigger order must be after primary JUDGE"
+  if ! grep -Fq 'Verify dual-judge is default-when-available' "$file" \
+    || ! grep -Fq 'persist `pair_trigger` before spawn with `pair.default`' "$file" \
+    || ! grep -Fq 'dispatch both judges concurrently against the same frozen diff' "$file" \
+    || ! grep -Fq 'A primary blocker never cancels the pair judge' "$file" \
+    || ! grep -Fq 'run the same two required judges sequentially' "$file" \
+    || ! grep -Fq 'Eligible schema-v3 state must include `pair.default`' "$file" \
+    || ! grep -Fq '`primary_judge_blocker` is parser-recognized only for archived v2.0 replay; new runs never write it' "$file"; then
+    bad "$file — VERIFY default pair rule missing."
     pair_trigger_order_missing=1
   fi
 done
@@ -1332,11 +1482,14 @@ for file in \
   .claude/skills/devlyn:resolve/references/phases/verify.md \
   .agents/skills/devlyn:resolve/references/phases/verify.md
 do
-  if ! grep -Fq 'Pair-mode is eligible only after MECHANICAL and the primary JUDGE have no' "$file" \
-    || ! grep -Fq 'After MECHANICAL and the primary JUDGE finish, compute and persist this before' "$file" \
-    || ! grep -Fq 'Malformed `state.risk_profile` is a VERIFY contract violation' "$file" \
-    || ! grep -Fq 'primary_judge_blocker' "$file"; then
-    bad "$file — VERIFY phase body must compute pair_trigger after primary JUDGE"
+  if ! grep -Fq 'Pair-mode (default when OTHER engine is available)' "$file" \
+    || ! grep -Fq 'with `pair.default` plus every applicable' "$file" \
+    || ! grep -Fq 'dispatch the primary and pair JUDGEs concurrently against the same frozen' "$file" \
+    || ! grep -Fq 'A primary blocker does not cancel the pair-JUDGE' "$file" \
+    || ! grep -Fq 'runs the same two required judges' "$file" \
+    || ! grep -Fq 'Eligible schema-v3 state must contain `pair.default`' "$file" \
+    || ! grep -Fq 'New runs never' "$file"; then
+    bad "$file — VERIFY default pair dispatch missing"
     pair_trigger_order_missing=1
   fi
 done
@@ -1345,16 +1498,50 @@ for file in \
   .claude/skills/devlyn:resolve/references/state-schema.md \
   .agents/skills/devlyn:resolve/references/state-schema.md
 do
-  if ! grep -Fq 'MECHANICAL and the primary JUDGE have no verdict-binding blockers' "$file" \
-    || ! grep -Fq 'may set only `user_no_pair`, `mechanical_blocker`, `primary_judge_blocker`, `auto_pair_other_engine_unavailable`, or null' "$file" \
-    || ! grep -Fq '`risk_profile` must remain an object with boolean' "$file" \
-    || ! grep -Fq 'state implies a pair decision is required but `pair_trigger` is missing' "$file"; then
-    bad "$file — state schema must document pair_trigger blocker and missing-trigger enforcement"
+  if ! grep -Fq '"version": "3.0"' "$file" \
+    || ! grep -Fq 'pair.default` dispatches both judges concurrently' "$file" \
+    || ! grep -Fq 'runs the same two required judges sequentially' "$file" \
+    || ! grep -Fq 'a primary blocker never skips the pair' "$file" \
+    || ! grep -Fq 'Schema v3.0 eligible state requires `pair.default`' "$file" \
+    || ! grep -Fq 'archived v2.0 replay' "$file" \
+    || ! grep -Fq 'Missing, contradictory, incomplete, or unknown trigger state BLOCKs VERIFY' "$file"; then
+    bad "$file — schema-v3 default pair contract missing"
     pair_trigger_order_missing=1
   fi
 done
 if [ $pair_trigger_order_missing -eq 0 ]; then
-  ok "VERIFY pair trigger order waits for primary JUDGE evidence"
+  ok "VERIFY default pair dispatches concurrently"
+fi
+
+section "Check 6k: PLAN route is orchestrator-fixed"
+plan_route_ok=1
+for plan_route_file in "config/skills/devlyn:resolve/SKILL.md" CLAUDE.md AGENTS.md \
+  "config/skills/devlyn:engines/SKILL.md" ".agents/skills/devlyn:resolve/SKILL.md" \
+  ".agents/skills/devlyn:engines/SKILL.md"; do
+  [ -f "$plan_route_file" ] || plan_route_ok=0
+done
+sed -n '/^<engine_routing>$/,/^<\/engine_routing>$/p' "config/skills/devlyn:resolve/SKILL.md" \
+  | grep -Fq 'PLAN is orchestrator-fixed and never inherits `--engine`, an executor pin, or `state.engine`' || plan_route_ok=0
+sed -n '/^## PHASE 1: PLAN$/,/^## PHASE 1.5/p' "config/skills/devlyn:resolve/SKILL.md" \
+  | grep -Fq 'PLAN is orchestrator-fixed and never inherits `--engine`, an executor pin, or `state.engine`' || plan_route_ok=0
+grep -Fq 'PLAN is orchestrator-fixed and never inherits `--engine` or an executor pin' CLAUDE.md || plan_route_ok=0
+grep -Fq 'PLAN is orchestrator-fixed and never inherits `--engine` or the executor pin' AGENTS.md || plan_route_ok=0
+grep -Fq 'PLAN is orchestrator-fixed and never follows the pin' config/skills/devlyn:engines/SKILL.md || plan_route_ok=0
+if grep -En '([Ee]xecutor|Default engine: Claude for)[^[:cntrl:]]{0,80}PLAN[[:space:]]*/[[:space:]]*IMPLEMENT([[:space:]]*/[[:space:]]*BUILD_GATE)?[[:space:]]*/[[:space:]]*CLEANUP' \
+    config/skills/devlyn:resolve/SKILL.md \
+    CLAUDE.md \
+    AGENTS.md \
+    README.md \
+    config/skills/devlyn:engines/SKILL.md \
+    .agents/skills/devlyn:resolve/SKILL.md \
+    .agents/skills/devlyn:engines/SKILL.md \
+    >/dev/null 2>&1; then
+  plan_route_ok=0
+fi
+if [ $plan_route_ok -eq 1 ]; then
+  ok "PLAN is orchestrator-fixed: exact declarations locked on all surfaces, no executor-role enumeration"
+else
+  bad "PLAN route contract violated: a locked orchestrator-fixed declaration is missing or an executor-role enumeration re-teaches PLAN"
 fi
 
 section "Check 6h: No undocumented spec.expected.json.browser_flows field"
@@ -3444,8 +3631,9 @@ else
   offenders="${offenders}"$'\n'"packaged pair evidence subset could not allocate temporary audit workspace"
 fi
 if make_temp_file pack_json /tmp/devlyn-lint-pack.XXXXXX.json \
-  && npm pack --dry-run --json > "$pack_json" 2>/dev/null; then
-  if ! node - "$pack_json" <<'NODE'
+  && make_temp_dir pack_cache /tmp/devlyn-lint-pack-cache.XXXXXX; then
+  if npm_config_cache="$pack_cache" npm pack --dry-run --json > "$pack_json" 2>/dev/null; then
+    if ! node - "$pack_json" <<'NODE'
 const fs = require("fs");
 const path = require("path");
 const packPath = process.argv[2];
@@ -3531,13 +3719,17 @@ if (forbidden.length > 0) {
   process.exit(2);
 }
 NODE
-  then
-    offenders="${offenders}"$'\n'"npm pack dry-run must include benchmark runner/gate regression tests, all shadow fixture files, retired fixture replay docs, and exclude pycache artifacts"
+    then
+      offenders="${offenders}"$'\n'"npm pack dry-run must include benchmark runner/gate regression tests, all shadow fixture files, retired fixture replay docs, and exclude pycache artifacts"
+    fi
+  else
+    offenders="${offenders}"$'\n'"npm pack dry-run failed while checking benchmark runner/gate regression tests"
   fi
 else
-  offenders="${offenders}"$'\n'"npm pack dry-run failed while checking benchmark runner/gate regression tests"
+  offenders="${offenders}"$'\n'"npm pack dry-run could not allocate temporary package workspace"
 fi
 [ -n "${pack_json:-}" ] && rm -f "$pack_json"
+[ -n "${pack_cache:-}" ] && rm -rf "$pack_cache"
 non_executable_shell_scripts=$(find benchmark/auto-resolve/scripts -maxdepth 1 -name '*.sh' -type f ! -perm -111 -print | sort)
 if [ -n "$non_executable_shell_scripts" ]; then
   while IFS= read -r f; do
@@ -3694,7 +3886,7 @@ if ! grep -Fq 'high-risk fixture must include a resolve risk-trigger term' scrip
   || ! grep -Fq 'expected.json must be an object' scripts/lint-fixtures.sh \
   || ! grep -Fq 'schema_ok=0' scripts/lint-fixtures.sh \
   || ! grep -Fq 'DEVLYN_LINT_FIXTURES_NO_JSONSCHEMA' scripts/lint-fixtures.sh \
-  || ! grep -Fq 'conditional pair/risk-probe triggers' benchmark/auto-resolve/fixtures/SCHEMA.md; then
+  || ! grep -Fq 'conditional risk-probe triggers' benchmark/auto-resolve/fixtures/SCHEMA.md; then
   offenders="${offenders}"$'\n'"fixture lint must require high-risk fixtures to include resolve pair/risk trigger terms"
 fi
 unsafe_json_parser_refs=$(python3 - <<'PY'