devlyn-cli 2.3.0 → 2.3.2

package/benchmark/auto-resolve/scripts/recent-benchmark-summary.py

@@ -0,0 +1,232 @@
+#!/usr/bin/env python3
+"""Print a compact, wrap-safe benchmark snapshot from local artifacts."""
+from __future__ import annotations
+
+import argparse
+import importlib.util
+import json
+import pathlib
+import sys
+import textwrap
+from typing import Any
+
+SCRIPT_DIR = pathlib.Path(__file__).resolve().parent
+FRONTIER_PATH = SCRIPT_DIR / "pair-candidate-frontier.py"
+
+
+def load_frontier_module() -> Any:
+    spec = importlib.util.spec_from_file_location("pair_candidate_frontier", FRONTIER_PATH)
+    if spec is None or spec.loader is None:
+        raise RuntimeError(f"cannot load frontier module: {FRONTIER_PATH}")
+    module = importlib.util.module_from_spec(spec)
+    spec.loader.exec_module(module)
+    return module
+
+
+FRONTIER = load_frontier_module()
+
+
+def best_rows(report: dict[str, Any]) -> list[dict[str, Any]]:
+    rows: list[dict[str, Any]] = []
+    for row in report.get("rows", []):
+        if row.get("status") != "pair_evidence_passed":
+            continue
+        best = FRONTIER.best_pair_evidence(row.get("passing_pair_evidence", []))
+        if best is None:
+            continue
+        rows.append({"fixture": row["fixture"], **best})
+    return rows
+
+
+def display_fixture(fixture: str) -> str:
+    short, _, rest = fixture.partition("-")
+    return f"{short} {rest.replace('-', ' ')}" if rest else fixture
+
+
+def fmt_margin(value: Any) -> str:
+    return f"{value:+d}" if isinstance(value, int) and not isinstance(value, bool) else "n/a"
+
+
+def fmt_decimal_margin(value: Any) -> str:
+    return f"{value:+.2f}" if isinstance(value, (int, float)) and not isinstance(value, bool) else "n/a"
+
+
+def fmt_wall(value: Any) -> str:
+    return f"{value:.2f}x" if isinstance(value, (int, float)) and not isinstance(value, bool) else "n/a"
+
+
+def fmt_score(value: Any) -> str:
+    return str(value) if isinstance(value, int) and not isinstance(value, bool) else "n/a"
+
+
+def wrap_item(prefix: str, text: str, *, width: int) -> list[str]:
+    return textwrap.wrap(
+        text,
+        width=width,
+        initial_indent=prefix,
+        subsequent_indent=" " * len(prefix),
+        break_long_words=False,
+        break_on_hyphens=False,
+    ) or [prefix.rstrip()]
+
+
+def render_text(report: dict[str, Any], *, width: int) -> str:
+    rows = best_rows(report)
+    lines = [
+        "Recent Benchmark Snapshot",
+        "=========================",
+        "",
+        "Status",
+        f"  Verdict: {report.get('verdict', 'n/a')}",
+        f"  Active fixtures: {report.get('fixtures_total', 'n/a')}",
+        f"  Rejected controls: {report.get('rejected_total', 'n/a')}",
+        f"  Pair evidence rows: {report.get('pair_evidence_total', 'n/a')}",
+        f"  Unmeasured candidates: {report.get('unmeasured_candidate_total', 'n/a')}",
+        "",
+        "Pair Lift",
+        f"  Average margin: {fmt_decimal_margin(report.get('pair_margin_avg'))}",
+        f"  Minimum margin: {fmt_margin(report.get('pair_margin_min'))}",
+        f"  Average wall ratio: {fmt_wall(report.get('pair_solo_wall_ratio_avg'))}",
+        f"  Maximum wall ratio: {fmt_wall(report.get('pair_solo_wall_ratio_max'))}",
+        f"  Gate: margin >= {fmt_margin(report.get('min_pair_margin'))}; wall <= {fmt_wall(report.get('max_pair_solo_wall_ratio'))}",
+        "",
+        "Pair Evidence",
+    ]
+    if not rows:
+        lines.append("  No passing pair evidence rows found.")
+        return "\n".join(lines) + "\n"
+
+    for item in rows:
+        lines.append(f"  {display_fixture(item['fixture'])}")
+        lines.append(
+            "    scores: bare {bare} | solo_claude {solo} | pair {pair}".format(
+                bare=fmt_score(item.get("bare_score")),
+                solo=fmt_score(item.get("solo_score")),
+                pair=fmt_score(item.get("pair_score")),
+            )
+        )
+        lines.append(
+            "    lift: {margin} | wall {wall} | arm {arm}".format(
+                margin=fmt_margin(item.get("pair_margin")),
+                wall=fmt_wall(item.get("pair_solo_wall_ratio")),
+                arm=item.get("pair_arm") or "n/a",
+            )
+        )
+        lines.extend(wrap_item("    run: ", str(item.get("run_id") or "n/a"), width=width))
+        triggers = ", ".join(item.get("pair_trigger_reasons") or [])
+        lines.extend(wrap_item("    triggers: ", triggers or "n/a", width=width))
+    return "\n".join(lines) + "\n"
+
+
+def render_markdown(report: dict[str, Any], *, width: int) -> str:
+    rows = best_rows(report)
+    lines = [
+        "# Recent Benchmark Snapshot",
+        "",
+        "## Status",
+        "",
+        f"- Verdict: **{report.get('verdict', 'n/a')}**",
+        f"- Active fixtures: {report.get('fixtures_total', 'n/a')}",
+        f"- Rejected controls: {report.get('rejected_total', 'n/a')}",
+        f"- Pair evidence rows: {report.get('pair_evidence_total', 'n/a')}",
+        f"- Unmeasured candidates: {report.get('unmeasured_candidate_total', 'n/a')}",
+        "",
+        "## Pair Lift",
+        "",
+        f"- Average margin: **{fmt_decimal_margin(report.get('pair_margin_avg'))}**",
+        f"- Minimum margin: **{fmt_margin(report.get('pair_margin_min'))}**",
+        f"- Average wall ratio: {fmt_wall(report.get('pair_solo_wall_ratio_avg'))}",
+        f"- Maximum wall ratio: {fmt_wall(report.get('pair_solo_wall_ratio_max'))}",
+        f"- Gate: margin >= {fmt_margin(report.get('min_pair_margin'))}; wall <= {fmt_wall(report.get('max_pair_solo_wall_ratio'))}",
+        "",
+        "## Pair Evidence",
+        "",
+    ]
+    if not rows:
+        lines.append("No passing pair evidence rows found.")
+        return "\n".join(lines) + "\n"
+
+    for item in rows:
+        lines.extend(
+            [
+                f"### {display_fixture(item['fixture'])}",
+                "",
+                f"- Scores: bare {fmt_score(item.get('bare_score'))}, solo_claude {fmt_score(item.get('solo_score'))}, pair {fmt_score(item.get('pair_score'))}.",
+                f"- Lift: {fmt_margin(item.get('pair_margin'))}; wall {fmt_wall(item.get('pair_solo_wall_ratio'))}; arm `{item.get('pair_arm') or 'n/a'}`.",
+                f"- Run: `{item.get('run_id') or 'n/a'}`.",
+            ]
+        )
+        triggers = ", ".join(item.get("pair_trigger_reasons") or [])
+        wrapped = wrap_item("- Triggers: ", triggers or "n/a", width=width)
+        lines.extend(wrapped)
+        lines.append("")
+    return "\n".join(lines).rstrip() + "\n"
+
+
+def main() -> int:
+    parser = argparse.ArgumentParser()
+    parser.add_argument(
+        "--fixtures-root",
+        type=pathlib.Path,
+        default=pathlib.Path("benchmark/auto-resolve/fixtures"),
+    )
+    parser.add_argument(
+        "--registry",
+        type=pathlib.Path,
+        default=SCRIPT_DIR / "pair-rejected-fixtures.sh",
+    )
+    parser.add_argument(
+        "--results-root",
+        type=pathlib.Path,
+        default=pathlib.Path("benchmark/auto-resolve/results"),
+    )
+    parser.add_argument("--out-json", type=pathlib.Path)
+    parser.add_argument("--out-md", type=pathlib.Path)
+    parser.add_argument(
+        "--max-width",
+        type=int,
+        default=92,
+        help="target maximum line width for text and markdown output",
+    )
+    parser.add_argument(
+        "--min-pair-margin",
+        type=int,
+        default=5,
+        help="minimum pair-over-solo margin required to count passing pair evidence",
+    )
+    parser.add_argument(
+        "--max-pair-solo-wall-ratio",
+        type=float,
+        default=3.0,
+        help="maximum pair/solo wall-time ratio allowed to count passing pair evidence",
+    )
+    args = parser.parse_args()
+    if args.max_width < 60:
+        print("error: --max-width must be >= 60", file=sys.stderr)
+        return 2
+
+    try:
+        report = FRONTIER.build_report(
+            fixtures_root=args.fixtures_root,
+            registry=args.registry,
+            results_root=args.results_root,
+            min_pair_margin=args.min_pair_margin,
+            max_pair_solo_wall_ratio=args.max_pair_solo_wall_ratio,
+        )
+    except ValueError as exc:
+        print(f"error: {exc}", file=sys.stderr)
+        return 2
+
+    if args.out_json:
+        args.out_json.parent.mkdir(parents=True, exist_ok=True)
+        args.out_json.write_text(json.dumps(report, indent=2) + "\n", encoding="utf8")
+    if args.out_md:
+        args.out_md.parent.mkdir(parents=True, exist_ok=True)
+        args.out_md.write_text(render_markdown(report, width=args.max_width), encoding="utf8")
+
+    print(render_text(report, width=args.max_width), end="")
+    return 0
+
+
+if __name__ == "__main__":
+    sys.exit(main())

package/benchmark/auto-resolve/scripts/run-fixture.sh

@@ -5,8 +5,8 @@
 # subprocess (isolated session), then captures artifacts + runs verification.
 #
 # Usage:
-#   run-fixture.sh --fixture <FID> --arm <variant|bare> --run-id <ID>
-#   run-fixture.sh --fixture <FID> --arm <variant|bare> --run-id <ID> --dry-run
+#   run-fixture.sh --fixture <FID> --arm <variant|solo_claude|bare|l2_gated|l2_risk_probes|l2_forced> --run-id <ID>
+#   run-fixture.sh --fixture <FID> --arm <variant|solo_claude|bare|l2_gated|l2_risk_probes|l2_forced> --run-id <ID> --dry-run
 #
 # Outputs to benchmark/auto-resolve/results/<run-id>/<fixture>/<arm>/:
 #   input.md, transcript.txt, diff.patch, changed-files.txt, verify.json,
@@ -19,6 +19,15 @@ usage() {
   exit 1
 }
 
+require_value() {
+  local flag="$1"
+  local value="${2:-}"
+  if [ -z "$value" ] || [[ "$value" == --* ]]; then
+    echo "$flag requires a value" >&2
+    exit 1
+  fi
+}
+
 kill_worktree_processes() {
   local work_dir="$1"
   local signal="$2"
@@ -40,16 +49,16 @@ FIXTURE=""; ARM=""; RUN_ID=""; DRY_RUN=0
 RESOLVE_SKILL="new"
 while [ $# -gt 0 ]; do
   case "$1" in
-    --fixture)        FIXTURE="$2"; shift 2;;
-    --arm)            ARM="$2";     shift 2;;
-    --run-id)         RUN_ID="$2";  shift 2;;
-    --resolve-skill)  RESOLVE_SKILL="$2"; shift 2;;
+    --fixture)        require_value "$1" "${2:-}"; FIXTURE="$2"; shift 2;;
+    --arm)            require_value "$1" "${2:-}"; ARM="$2";     shift 2;;
+    --run-id)         require_value "$1" "${2:-}"; RUN_ID="$2";  shift 2;;
+    --resolve-skill)  require_value "$1" "${2:-}"; RESOLVE_SKILL="$2"; shift 2;;
     --dry-run)        DRY_RUN=1;    shift;;
     *) usage;;
   esac
 done
 [ -n "$FIXTURE" ] && [ -n "$ARM" ] && [ -n "$RUN_ID" ] || usage
-# iter-0019: original 3 arms — variant (L2-old: Claude orchestrator + Codex BUILD pair via --engine auto),
+# iter-0019/0037: 3 smoke arms — variant (L2: Claude orchestrator + risk-probes pair path),
 # solo_claude (L1: Claude orchestrator, codex blocked by shim+wrapper enforcement),
 # bare (L0: direct claude -p, no skill, no codex).
 # iter-0033c (Codex R0-infra adoption, 2026-05-02): two L2 diagnostic arms for /devlyn:resolve —
@@ -99,8 +108,21 @@ for f in "$META" "$EXPECTED" "$SPEC" "$TASK"; do
   [ -f "$f" ] || { echo "fixture missing required file: $f (see SCHEMA.md)"; exit 1; }
 done
 
-TIMEOUT=$(python3 -c "import json; print(json.load(open('$META'))['timeout_seconds'])")
-if [ "$ARM" = "l2_risk_probes" ]; then
+TIMEOUT=$(python3 - "$META" "$BENCH_ROOT/scripts" <<'PY'
+import pathlib
+import sys
+
+sys.path.insert(0, sys.argv[2])
+from pair_evidence_contract import loads_strict_json_object
+
+metadata = loads_strict_json_object(pathlib.Path(sys.argv[1]).read_text())
+timeout = metadata.get("timeout_seconds")
+if not isinstance(timeout, int) or isinstance(timeout, bool) or timeout <= 0:
+    raise SystemExit("metadata timeout_seconds must be a positive integer")
+print(timeout)
+PY
+)
+if [ "$ARM" = "variant" ] || [ "$ARM" = "l2_risk_probes" ]; then
   # This arm adds a bounded Codex probe-derive phase before IMPLEMENT and a
   # bounded Codex pair-JUDGE during VERIFY. The full-pipeline gate still
   # enforces wall-time efficiency by pair/solo ratio; this budget prevents a
@@ -119,19 +141,18 @@ WORK_DIR="/tmp/bench-${RUN_ID}-${FIXTURE}-${ARM}"
 rm -rf "$WORK_DIR"
 cp -R "$BENCH_ROOT/fixtures/test-repo" "$WORK_DIR"
 
-# All skill-driven arms (variant / solo_claude / l2_gated / l2_forced) get
-# devlyn skills + project CLAUDE.md pre-baseline + codex shim + monitored
-# wrapper. Bare gets nothing (no skill, no shim, no env).
+# All skill-driven arms (variant / solo_claude / l2_gated / l2_risk_probes /
+# l2_forced) get devlyn skills + project CLAUDE.md pre-baseline + codex shim
+# + monitored wrapper. Bare gets nothing (no skill, no shim, no env).
 #
 # iter-0019: solo_claude (L1) shares variant-arm staging because the L1 arm
 # runs the same orchestrator on the same skills — only difference is codex
 # is blocked. Shim catches PATH resolution; wrapper catches direct-path
 # invocations.
-# iter-0033c (Codex R0-infra Q6): l2_gated/l2_forced share variant staging
-# (codex unblocked, shim+wrapper routing). Difference vs variant is the
-# ENGINE_CLAUSE branch below — l2_* run --engine claude (Claude IMPLEMENT)
-# while variant uses --engine auto (Codex IMPLEMENT). Pair-mode in
-# /devlyn:resolve VERIFY phase pulls Codex via the OTHER-engine rule.
+# iter-0033c/0037 (Codex R0-infra Q6 + risk probes): pair arms share variant
+# staging (codex unblocked, shim+wrapper routing). The smoke `variant` arm now
+# follows the current measured risk-probes path rather than an older
+# auto-engine implement route.
 if [ "$ARM" = "variant" ] || [ "$ARM" = "solo_claude" ] \
    || [ "$ARM" = "l2_gated" ] || [ "$ARM" = "l2_risk_probes" ] || [ "$ARM" = "l2_forced" ]; then
   mkdir -p "$WORK_DIR/.claude"
@@ -183,7 +204,7 @@ if [ "$ARM" = "variant" ] || [ "$ARM" = "solo_claude" ] \
     # the bypass weapon. Across iter-0019 paid 5-fixture run the bypass
     # was OPEN but never exercised; this closes the surface preemptively
     # before iter-0020's 9-fixture L0/L1/L2 run.
-    # iter-0033c (Codex R0-infra Q5): l2_gated/l2_forced are codex-UNBLOCKED
+    # iter-0033c/0037 (Codex R0-infra Q5 + risk probes): l2_* arms are codex-UNBLOCKED
     # (codex must be reachable for VERIFY pair-JUDGE). They take the variant
     # path: ARM_CODEX_BLOCKED=0 → python writer omits CODEX_BLOCKED from env
     # entirely (the shim refuses on any non-empty value, so 0 ≠ unset).
@@ -209,11 +230,12 @@ if codex_blocked == "1":
     # CODEX_BLOCKED enforcement gap.
     env["CODEX_BLOCKED"] = "1"
 else:
-    # variant arm (L2) — codex routes through wrapper as part of pair-mode
-    # BUILD; both vars are required by the shim/wrapper handshake.
+    # variant / pair arms — codex routes through wrapper for risk-probe
+    # derivation and VERIFY pair-JUDGE; both vars are required by the
+    # shim/wrapper handshake.
     env["CODEX_REAL_BIN"] = real_bin
     env["CODEX_MONITORED_PATH"] = monitored
-    if arm == "l2_risk_probes":
+    if arm in ("variant", "l2_risk_probes"):
         # Risk-probe derivation is a bounded contract-conversion step. A long
         # Codex run is a harness failure, not useful extra quality signal.
         env["CODEX_MONITORED_TIMEOUT_SEC"] = "300"
@@ -273,9 +295,12 @@ fi
 # files. Those commands still run in the post-run verifier below.
 if [ "$ARM" = "variant" ] || [ "$ARM" = "solo_claude" ] \
    || [ "$ARM" = "l2_gated" ] || [ "$ARM" = "l2_risk_probes" ] || [ "$ARM" = "l2_forced" ]; then
-  python3 - "$EXPECTED" "$WORK_DIR/.devlyn/spec-verify.json" <<'PY'
-import json, os, sys
-expected = json.load(open(sys.argv[1]))
+  python3 - "$EXPECTED" "$WORK_DIR/.devlyn/spec-verify.json" "$BENCH_ROOT/scripts" <<'PY'
+import json, os, pathlib, sys
+sys.path.insert(0, sys.argv[3])
+from pair_evidence_contract import loads_strict_json_object
+
+expected = loads_strict_json_object(pathlib.Path(sys.argv[1]).read_text())
 out_path = sys.argv[2]
 visible_commands = [
     cmd for cmd in expected.get("verification_commands", [])
@@ -301,11 +326,11 @@ fi
 #   2. Spec-mode `/devlyn:resolve --spec <path>` for the rest (post iter-0034
 #      Phase 4 cutover the OLD `/devlyn:auto-resolve` route was deleted).
 PROMPT_FILE="$RESULT_DIR/input.md"
-# Variant uses --engine auto (experimental dual-engine: codex BUILD + claude
-# critique pair); solo_claude uses --engine claude explicitly so the orchestrator
-# routes every phase to Claude and never tries to invoke codex. The CODEX_BLOCKED
-# shim enforces this at the binary layer if the orchestrator misroutes. Both
-# arms pass the engine flag explicitly so they survive future runtime-default
+# Variant uses the current measured risk-probes pair path; solo_claude uses
+# --engine claude explicitly so the orchestrator routes every implementation
+# phase to Claude and never tries to invoke codex. The CODEX_BLOCKED shim
+# enforces this at the binary layer if the orchestrator misroutes. Both arms
+# pass the engine flag explicitly so they survive future runtime-default
 # changes (post iter-0020 close-out: default flipped to claude).
 if [ "$ARM" = "variant" ] || [ "$ARM" = "solo_claude" ] \
    || [ "$ARM" = "l2_gated" ] || [ "$ARM" = "l2_risk_probes" ] || [ "$ARM" = "l2_forced" ]; then
@@ -315,8 +340,8 @@ if [ "$ARM" = "variant" ] || [ "$ARM" = "solo_claude" ] \
       ENGINE_PROMPT_HINT="Run with \`--engine claude\` for every phase. Codex must not be invoked — the harness has blocked it at the binary layer for this run."
       ;;
     variant)
-      ENGINE_CLAUSE="--engine auto"
-      ENGINE_PROMPT_HINT="Run with \`--engine auto\` so the experimental dual-engine routing fires (Codex BUILD/FIX, Claude EVAL/CRITIC) — do not override it."
+      ENGINE_CLAUSE="--engine claude --risk-probes"
+      ENGINE_PROMPT_HINT="Run with \`--engine claude --risk-probes\` so the smoke L2 arm uses the current measured pair path: Claude implements, Codex derives bounded visible-verification probes and can act as VERIFY pair-JUDGE."
       ;;
     l2_gated)
       # NEW L2 with natural pair-mode triggers. Claude does IMPLEMENT;
@@ -484,7 +509,7 @@ else
     # iter-0009 + iter-0019: prepend codex shim PATH for any arm that staged
     # one. variant routes through codex-monitored.sh; solo_claude refuses on
     # CODEX_BLOCKED=1; bare has no shim.
-    # iter-0033c (Codex R0-infra Q6): l2_gated/l2_forced ALSO need the shim
+    # iter-0033c/0037 (Codex R0-infra Q6 + risk probes): l2_* arms ALSO need the shim
     # PATH — they route Claude IMPLEMENT but Codex pair-JUDGE in VERIFY hits
     # `codex exec` through the wrapper for starvation safety.
     if { [ "$ARM" = "variant" ] || [ "$ARM" = "solo_claude" ] \
@@ -652,10 +677,13 @@ fi
 # Run verification commands + forbidden pattern scan + deps check. Uses
 # the operator's real HOME (same as the arm saw). Fixtures that need HOME
 # isolation override it inline per verification command.
-python3 - "$EXPECTED" "$RESULT_DIR" "$WORK_DIR" <<'PY'
-import json, os, re, subprocess, sys
+python3 - "$EXPECTED" "$RESULT_DIR" "$WORK_DIR" "$BENCH_ROOT/scripts" <<'PY'
+import json, os, pathlib, re, subprocess, sys
+
+sys.path.insert(0, sys.argv[4])
+from pair_evidence_contract import loads_strict_json_object
 
-expected = json.load(open(sys.argv[1]))
+expected = loads_strict_json_object(pathlib.Path(sys.argv[1]).read_text())
 result_dir = sys.argv[2]
 work = sys.argv[3]
 
@@ -771,12 +799,39 @@ for oracle_file in (
     "oracle-scope-tier-b.json",
     "oracle-test-fidelity.json",
 ):
+    oracle_path = os.path.join(result_dir, oracle_file)
     try:
-        data = json.load(open(os.path.join(result_dir, oracle_file)))
-    except Exception:
+        raw_oracle = loads_strict_json_object(pathlib.Path(oracle_path).read_text())
+    except (OSError, ValueError) as exc:
+        oracle_name = oracle_file.removesuffix(".json")
+        verify["oracle_findings"].append({
+            "oracle": oracle_name,
+            "type": "oracle-error",
+            "severity": "hard",
+            "verdict": "Deterministic oracle failed or emitted an invalid artifact",
+            "error": f"oracle artifact malformed or unreadable: {exc}",
+        })
+        verify["oracle_disqualifier"] = True
         continue
+    data = raw_oracle
     oracle_name = data.get("oracle") or oracle_file.removesuffix(".json")
-    for finding in data.get("findings", []) or []:
+    if not isinstance(oracle_name, str) or not oracle_name:
+        oracle_name = oracle_file.removesuffix(".json")
+    oracle_error = data.get("error")
+    if isinstance(oracle_error, str) and oracle_error:
+        verify["oracle_findings"].append({
+            "oracle": oracle_name,
+            "type": "oracle-error",
+            "severity": "hard",
+            "verdict": "Deterministic oracle failed or emitted an invalid artifact",
+            "error": oracle_error,
+        })
+        verify["oracle_disqualifier"] = True
+    raw_findings = data.get("findings")
+    findings = raw_findings if isinstance(raw_findings, list) else []
+    for finding in findings:
+        if not isinstance(finding, dict):
+            continue
         item = dict(finding)
         item["oracle"] = oracle_name
         verify["oracle_findings"].append(item)
@@ -796,12 +851,15 @@ PY
 
 # Timing + aggregate
 export INVOKE_EXIT WATCHDOG_FIRED
-python3 - "$RESULT_DIR" "$FIXTURE" "$ARM" "$RUN_ID" "$T_END" "$ELAPSED" "$TIMEOUT" <<'PY'
-import json, os, sys
+python3 - "$RESULT_DIR" "$FIXTURE" "$ARM" "$RUN_ID" "$T_END" "$ELAPSED" "$TIMEOUT" "$BENCH_ROOT/scripts" <<'PY'
+import json, os, pathlib, sys
 result_dir, fixture, arm, run_id = sys.argv[1:5]
 t_end, elapsed, timeout = int(sys.argv[5]), int(sys.argv[6]), int(sys.argv[7])
 
-timing = json.load(open(os.path.join(result_dir, "timing.json")))
+sys.path.insert(0, sys.argv[8])
+from pair_evidence_contract import loads_strict_json_object
+
+timing = loads_strict_json_object(pathlib.Path(result_dir, "timing.json").read_text())
 timing["end_epoch"] = t_end
 timing["elapsed_seconds"] = elapsed
 timing["timeout_seconds"] = timeout
@@ -812,7 +870,10 @@ timing["timeout_seconds"] = timeout
 timing["timed_out"] = os.environ.get("WATCHDOG_FIRED", "0") == "1"
 json.dump(timing, open(os.path.join(result_dir, "timing.json"), "w"), indent=2)
 
-verify = json.load(open(os.path.join(result_dir, "verify.json")))
+def as_dict(value):
+    return value if isinstance(value, dict) else {}
+
+verify = as_dict(loads_strict_json_object(pathlib.Path(result_dir, "verify.json").read_text()))
 try:
     with open(os.path.join(result_dir, "diff.patch")) as f: diff_size = len(f.read())
 except Exception: diff_size = 0
@@ -825,15 +886,21 @@ except Exception:
 state = {}
 state_path = os.path.join(result_dir, "run-archive", "pipeline.state.json")
 if os.path.isfile(state_path):
-    with open(state_path) as f:
-        state = json.load(f)
-verify_phase = (state.get("phases") or {}).get("verify") or {}
+    state = as_dict(loads_strict_json_object(pathlib.Path(state_path).read_text()))
+phases = as_dict(state.get("phases"))
+verify_phase = as_dict(phases.get("verify"))
+legacy_verify = as_dict(state.get("verify"))
 sub_verdicts = verify_phase.get("sub_verdicts")
-pair_trigger = verify_phase.get("pair_trigger") or ((state.get("verify") or {}).get("pair_trigger"))
-pair_mode = bool(
-    isinstance(sub_verdicts, dict)
-    and (sub_verdicts.get("judge_codex") is not None or sub_verdicts.get("pair_judge") is not None)
-) or bool(verify_phase.get("pair_mode"))
+pair_trigger = verify_phase.get("pair_trigger") or legacy_verify.get("pair_trigger")
+PAIR_VERDICTS = {"PASS", "PASS_WITH_ISSUES", "NEEDS_WORK", "BLOCKED", "FAIL"}
+
+def has_pair_judge_verdict(sub_verdicts):
+    return isinstance(sub_verdicts, dict) and (
+        sub_verdicts.get("judge_codex") in PAIR_VERDICTS
+        or sub_verdicts.get("pair_judge") in PAIR_VERDICTS
+    )
+
+pair_mode = has_pair_judge_verdict(sub_verdicts) or verify_phase.get("pair_mode") is True
 
 invoke_exit = int(os.environ.get("INVOKE_EXIT", "0"))
 plugin_contamination = False
@@ -893,7 +960,7 @@ result = {
     "invoke_exit": invoke_exit,
     "invoke_failure": invoke_failure,
     "invoke_failure_reason": invoke_failure_reason,
-    "terminal_verdict": ((state.get("phases") or {}).get("final_report") or {}).get("verdict"),
+    "terminal_verdict": as_dict(phases.get("final_report")).get("verdict"),
     "verify_verdict": verify_phase.get("verdict"),
     "pair_trigger": pair_trigger,
     "pair_mode": pair_mode,