Add signal-targeted intrinsic raw variant expansion (sprint 260)

2026-02-26 16:37:57 -07:00
parent ba1e0dc3e8
commit 42c53da5c6
8 changed files with 201 additions and 1 deletions
--- a/docs/constructive_editing_runtime_plan.md
+++ b/docs/constructive_editing_runtime_plan.md
@@ -232,3 +232,8 @@ Planning runtime controls (active):
  - `WSTONE_NATIVE_RAW_TOP_GAP_ADAPTIVE_RETRY`
  - `WSTONE_NATIVE_RAW_TOP_GAP_RETRY_SIGNALS`
  - `native_raw_adaptive_retry`
+- raw candidate search can expand intrinsic exploration with signal-targeted variants:
+  - `tools/mcp/synthesize_raw_signal_targeted_variants.py`
+  - `WSTONE_NATIVE_RAW_SIGNAL_TARGETED_VARIANTS`
+  - `WSTONE_NATIVE_RAW_SIGNAL_TARGETED_MAX_VARIANTS`
+  - `native_raw_signal_targeted_variants`
--- a/docs/generator_readiness_gap_registry_2026-02-26.md
+++ b/docs/generator_readiness_gap_registry_2026-02-26.md
@@ -47,7 +47,7 @@ This is the canonical dated registry for "not production-ready" generator gaps.
 | GR-018 | Performance/security/rollout constrained refactor enforcement gap | `docs/gap_hunt_fullstack_multifile_2026-02-26.md`, `logs/taskitem_runs/challenging_fullstack_multifile_20260226_r5/results.jsonl` | `partial` | Hard checks now enforce migration rollback + data-loss policy, security deny-by-default, SLO p95 presence, and rollout staged+abort policy. Enforcement is contract-level; generator capability under these constraints is still weak in C++ AB path. | Constraint-aware generation follow-up |
 | GR-019 | Parity-blocked readiness load (gating without capability closure) | `logs/taskitem_runs/challenging_fullstack_multifile_20260226_r7/summary.json`, `logs/taskitem_runs/challenging_subset_prod_20260226_r7/summary.json`, `docs/sprint225_227_execution_tracker_2026-02-26.md` | `partial` | Sprint 225-227 reduced blocked parity load from `6` to `0` on both tracked hard catalogs while keeping unresolved divergence at `0`. This closes immediate safety debt for current corpora, but robustness is still contingent on pattern-driven repair classes. | Generalize repairs beyond queue-shaped transpile outputs |
 | GR-020 | Semantic fallback overuse masks weak native decomposition | `logs/taskitem_runs/TEST_ONLY_sprint236_semantic_fallback_audit_20260226/semantic_fallback_summary.json`, `logs/taskitem_runs/TEST_ONLY_sprint237_semantic_fallback_gate_fail_20260226/semantic_fallback_budget_gate.json`, `logs/taskitem_runs/TEST_ONLY_sprint238_native_gate_fail_20260226.json`, `logs/taskitem_runs/TEST_ONLY_sprint239_semantic_fallback_audit_20260226/semantic_fallback_summary.json`, `logs/taskitem_runs/TEST_ONLY_sprint240_retry_20260226_145631/00_summary.json`, `docs/sprint236_execution_tracker_2026-02-26.md`, `docs/sprint237_execution_tracker_2026-02-26.md`, `docs/sprint238_execution_tracker_2026-02-26.md`, `docs/sprint239_execution_tracker_2026-02-26.md`, `docs/sprint240_execution_tracker_2026-02-26.md` | `partial` | Sprint 236 added deterministic fallback-gap auditing with tool + constraint metadata. Sprint 237 added fallback-budget hard gate (`max-fallback-rate`) and produced expected fail/pass artifacts. Sprint 238 added native decomposition hard gate in pipeline (`min task count`, `min semantic signal count`) so weak native generation can be blocked before fallback masking. Sprint 239 added native reason enrichment and improved semantic signal density (`0 -> 6`). Sprint 240 added native decomposition retry with explicit minimum-task policy; on current sample retry attempted but did not improve depth (`2 -> 2`). | Native decomposition task-depth upgrade (increase native task granularity beyond 2) |
-| GR-021 | Impact-specific native decomposition coverage not enforced uniformly | `docs/native_decomposition_impact_list_2026-02-26.md`, `tools/mcp/profiles/native_decomposition_impact_profiles.json`, `logs/taskitem_runs/TEST_ONLY_sprint241_fullstack_impact_20260226_150416/00_summary.json`, `logs/taskitem_runs/TEST_ONLY_sprint241_native_impact_aggregate_20260226/native_impact_coverage_aggregate.json`, `logs/taskitem_runs/TEST_ONLY_sprint242_impact_remediation_loop_20260226/remediation_loop_summary.json`, `logs/taskitem_runs/TEST_ONLY_sprint243_impact_remediation_tasks_loop_20260226_r2/remediation_loop_summary.json`, `logs/taskitem_runs/TEST_ONLY_sprint244_autofill_20260226_151651/00_summary.json`, `logs/taskitem_runs/TEST_ONLY_sprint244_autofill_enforce_20260226_151709/00_summary.json`, `logs/taskitem_runs/TEST_ONLY_sprint245_intrinsic_20260226_151835/00_summary.json`, `logs/taskitem_runs/TEST_ONLY_sprint246_multishot_20260226_153033/00_summary.json`, `logs/taskitem_runs/TEST_ONLY_sprint246_multishot_enforce_20260226_153045/00_summary.json`, `logs/taskitem_runs/TEST_ONLY_sprint247_singleshot_20260226_153230/00_summary.json`, `logs/taskitem_runs/TEST_ONLY_sprint247_singleshot_enforce_20260226_153241/00_summary.json`, `logs/taskitem_runs/TEST_ONLY_sprint248_rawsearch_20260226_153903/00_summary.json`, `logs/taskitem_runs/TEST_ONLY_sprint249_rawguard_20260226_154028/02ae_raw_candidate_search.json`, `logs/taskitem_runs/TEST_ONLY_sprint250_closure_ladder_20260226/closure_ladder_summary.json`, `logs/taskitem_runs/TEST_ONLY_sprint251_rawvariants_20260226_154355/00_summary.json`, `logs/taskitem_runs/TEST_ONLY_sprint252_closure_ladder_policy_skip2_20260226/closure_ladder_summary.json`, `logs/taskitem_runs/TEST_ONLY_sprint253_closure_ladder_batch_20260226/closure_ladder_batch_summary.json`, `logs/taskitem_runs/TEST_ONLY_sprint254_closure_ladder_batch_20260226_r2/raw_gap_backlog.json`, `logs/taskitem_runs/TEST_ONLY_sprint255_rawharden_20260226_161924/00_summary.json`, `logs/taskitem_runs/01a_fallback_intake_spec_20260226_162423/00_summary.json`, `logs/taskitem_runs/01a_fallback_intake_spec_20260226_162633/02ae_raw_candidate_search.json`, `logs/taskitem_runs/01a_fallback_intake_spec_20260226_162924/01e_raw_top_gap_requirements_report.json`, `logs/taskitem_runs/01a_fallback_intake_spec_20260226_163148/02af_raw_adaptive_retry_score.json` | `partial` | Sprint 241 introduced impact profiles/gates; 242-243 remediation loops; 244 first-pass autofill; 245 intrinsic constraints only (no uplift); 246 multishot closure; 247 single-shot shaping closure; 248 raw candidate search no uplift; 249 no-uplift guardrail; 250 closure ladder; 251 richer raw variants no uplift; 252 history-aware ladder routing; 253 batch ladder analytics; 254 raw gap backlog synthesis from `raw_only` attempts identifies top missing raw signals; 255 adds first-pass raw top-gap hardening that closes hard-sample profile coverage (`failing_profile_count 7 -> 0`) by injecting missing ops/contracts/reasons and minimum task depth; 256 adds top-gap weighted candidate selection telemetry and tie-break policy; 257 adds hard fail policy when top-gap uplift is required but absent (`exit 18`); 258 injects backlog-driven top-gap normalized requirements before raw generation; 259 adds adaptive retry with expanded top-gap requirements and deterministic retry scoring. Measured intrinsic output still shows no uplift on this sample (`retry_top_gap_score=0`, `retry_fail=1`, `applied=false`). | Add model/prompt-level intrinsic generation intervention keyed by missing-signal classes (beyond requirement text injection alone) |
+| GR-021 | Impact-specific native decomposition coverage not enforced uniformly | `docs/native_decomposition_impact_list_2026-02-26.md`, `tools/mcp/profiles/native_decomposition_impact_profiles.json`, `logs/taskitem_runs/TEST_ONLY_sprint241_fullstack_impact_20260226_150416/00_summary.json`, `logs/taskitem_runs/TEST_ONLY_sprint241_native_impact_aggregate_20260226/native_impact_coverage_aggregate.json`, `logs/taskitem_runs/TEST_ONLY_sprint242_impact_remediation_loop_20260226/remediation_loop_summary.json`, `logs/taskitem_runs/TEST_ONLY_sprint243_impact_remediation_tasks_loop_20260226_r2/remediation_loop_summary.json`, `logs/taskitem_runs/TEST_ONLY_sprint244_autofill_20260226_151651/00_summary.json`, `logs/taskitem_runs/TEST_ONLY_sprint244_autofill_enforce_20260226_151709/00_summary.json`, `logs/taskitem_runs/TEST_ONLY_sprint245_intrinsic_20260226_151835/00_summary.json`, `logs/taskitem_runs/TEST_ONLY_sprint246_multishot_20260226_153033/00_summary.json`, `logs/taskitem_runs/TEST_ONLY_sprint246_multishot_enforce_20260226_153045/00_summary.json`, `logs/taskitem_runs/TEST_ONLY_sprint247_singleshot_20260226_153230/00_summary.json`, `logs/taskitem_runs/TEST_ONLY_sprint247_singleshot_enforce_20260226_153241/00_summary.json`, `logs/taskitem_runs/TEST_ONLY_sprint248_rawsearch_20260226_153903/00_summary.json`, `logs/taskitem_runs/TEST_ONLY_sprint249_rawguard_20260226_154028/02ae_raw_candidate_search.json`, `logs/taskitem_runs/TEST_ONLY_sprint250_closure_ladder_20260226/closure_ladder_summary.json`, `logs/taskitem_runs/TEST_ONLY_sprint251_rawvariants_20260226_154355/00_summary.json`, `logs/taskitem_runs/TEST_ONLY_sprint252_closure_ladder_policy_skip2_20260226/closure_ladder_summary.json`, `logs/taskitem_runs/TEST_ONLY_sprint253_closure_ladder_batch_20260226/closure_ladder_batch_summary.json`, `logs/taskitem_runs/TEST_ONLY_sprint254_closure_ladder_batch_20260226_r2/raw_gap_backlog.json`, `logs/taskitem_runs/TEST_ONLY_sprint255_rawharden_20260226_161924/00_summary.json`, `logs/taskitem_runs/01a_fallback_intake_spec_20260226_162423/00_summary.json`, `logs/taskitem_runs/01a_fallback_intake_spec_20260226_162633/02ae_raw_candidate_search.json`, `logs/taskitem_runs/01a_fallback_intake_spec_20260226_162924/01e_raw_top_gap_requirements_report.json`, `logs/taskitem_runs/01a_fallback_intake_spec_20260226_163148/02af_raw_adaptive_retry_score.json`, `logs/taskitem_runs/01a_fallback_intake_spec_20260226_163556/02ae_raw_signal_targeted_variants.json` | `partial` | Sprint 241 introduced impact profiles/gates; 242-243 remediation loops; 244 first-pass autofill; 245 intrinsic constraints only (no uplift); 246 multishot closure; 247 single-shot shaping closure; 248 raw candidate search no uplift; 249 no-uplift guardrail; 250 closure ladder; 251 richer raw variants no uplift; 252 history-aware ladder routing; 253 batch ladder analytics; 254 raw gap backlog synthesis from `raw_only` attempts identifies top missing raw signals; 255 adds first-pass raw top-gap hardening that closes hard-sample profile coverage (`failing_profile_count 7 -> 0`) by injecting missing ops/contracts/reasons and minimum task depth; 256 adds top-gap weighted candidate selection telemetry and tie-break policy; 257 adds hard fail policy when top-gap uplift is required but absent (`exit 18`); 258 injects backlog-driven top-gap normalized requirements before raw generation; 259 adds adaptive retry with expanded top-gap requirements and deterministic retry scoring; 260 adds signal-targeted variant expansion keyed to missing-signal classes and increases intrinsic search breadth (`available_variants 2 -> 8`). Measured intrinsic output still shows no uplift on this sample (`best_top_gap_score=0`, `failing_profile_count=1`). | Add model/prompt-level intrinsic generation intervention keyed by missing-signal classes (beyond requirement text injection alone) |

 ## What Was Covered Today (Sprints 175-184)

--- a/docs/progress_log_2026-02-26.md
+++ b/docs/progress_log_2026-02-26.md
@@ -474,3 +474,19 @@
 - Current measured result on hard sample:
  - retry attempted and succeeded operationally
  - no intrinsic uplift (`retry_top_gap_score=0`, `retry_fail=1`, `applied=false`).
+
+## Sprint 260 Added (Same Day)
+
+- Added signal-targeted intrinsic variant expansion:
+  - `tools/mcp/synthesize_raw_signal_targeted_variants.py`
+  - pipeline controls:
+    - `WSTONE_NATIVE_RAW_SIGNAL_TARGETED_VARIANTS`
+    - `WSTONE_NATIVE_RAW_SIGNAL_TARGETED_MAX_VARIANTS`
+  - summary packet:
+    - `native_raw_signal_targeted_variants`
+- Dated A/B artifacts:
+  - OFF: `logs/taskitem_runs/01a_fallback_intake_spec_20260226_163629/00_summary.json`
+  - ON: `logs/taskitem_runs/01a_fallback_intake_spec_20260226_163556/00_summary.json`
+- Current measured result on hard sample:
+  - search breadth increased (`available_variants 2 -> 8`)
+  - intrinsic outcome unchanged (`selected_variant=0`, `best_top_gap_score=0`, `failing_profile_count=1`).
--- a/docs/sprint260_execution_tracker_2026-02-26.md
+++ b/docs/sprint260_execution_tracker_2026-02-26.md
@@ -0,0 +1,40 @@
+# Sprint 260 Execution Tracker - 2026-02-26
+
+## Scope
+- `sprint260_plan.md`
+
+## Implemented
+
+New tool:
+- `tools/mcp/synthesize_raw_signal_targeted_variants.py`
+  - Reads prioritized missing signals from raw backlog.
+  - Emits focused and bundled normalized-requirement variants keyed to signal class.
+
+Pipeline integration in `tools/mcp/run_sprint_taskitem_pipeline.sh`:
+- Added:
+  - `WSTONE_NATIVE_RAW_SIGNAL_TARGETED_VARIANTS`
+  - `WSTONE_NATIVE_RAW_SIGNAL_TARGETED_MAX_VARIANTS`
+- Raw search now optionally merges:
+  - profile variants (`synthesize_raw_candidate_requirement_variants.py`)
+  - signal-targeted variants (`synthesize_raw_signal_targeted_variants.py`)
+- Added summary packet:
+  - `native_raw_signal_targeted_variants`
+
+## Validation Artifacts
+
+OFF:
+- `logs/taskitem_runs/01a_fallback_intake_spec_20260226_163629/00_summary.json`
+
+ON:
+- `logs/taskitem_runs/01a_fallback_intake_spec_20260226_163556/00_summary.json`
+- generated signal variant payload:
+  - `logs/taskitem_runs/01a_fallback_intake_spec_20260226_163556/02ae_raw_signal_targeted_variants.json`
+
+Observed result on hard sample:
+- OFF: `available_variants=2`, `attempted_variants=3`, `selected_variant=0`
+- ON: `available_variants=8`, `attempted_variants=8`, `selected_variant=0`
+- `best_top_gap_score` remained `0` and `failing_profile_count` remained `1`
+
+## Explicit Completion Signal
+
+- Sprint 260: `DONE` (implemented + validated; intrinsic uplift still pending)
--- a/editor/src/Sprint260IntegrationSummary.h
+++ b/editor/src/Sprint260IntegrationSummary.h
@@ -0,0 +1,6 @@
+#pragma once
+
+// Sprint 260 integration summary:
+// - Added signal-targeted raw variant synthesis from prioritized missing backlog classes.
+// - Raw candidate search can merge profile variants with signal-targeted variants.
+// - Summary now exposes native_raw_signal_targeted_variants telemetry for intrinsic search analysis.
--- a/sprint260_plan.md
+++ b/sprint260_plan.md
@@ -0,0 +1,12 @@
+# Sprint 260 Plan: Signal-Targeted Intrinsic Variant Expansion
+
+## Goal
+Add missing-signal-class-targeted raw requirement variants so intrinsic candidate search explores direct interventions keyed to backlog failure classes.
+
+## Steps
+- Step 2347: Add tool to synthesize raw variants from top missing-signal backlog classes.
+- Step 2348: Merge signal-targeted variants with profile-based variants in raw search.
+- Step 2349: Add pipeline controls for enabling and sizing signal-targeted variant expansion.
+- Step 2350: Emit signal-targeted variant telemetry in summary packet.
+- Step 2351: Validate OFF/ON behavior on hard sample.
+- Step 2352: Add `Sprint260IntegrationSummary.h` and execution tracker.
--- a/tools/mcp/run_sprint_taskitem_pipeline.sh
+++ b/tools/mcp/run_sprint_taskitem_pipeline.sh
@@ -53,6 +53,8 @@ NATIVE_RAW_CANDIDATE_SEARCH="${WSTONE_NATIVE_RAW_CANDIDATE_SEARCH:-0}"
 NATIVE_RAW_CANDIDATE_MAX_VARIANTS="${WSTONE_NATIVE_RAW_CANDIDATE_MAX_VARIANTS:-8}"
 NATIVE_RAW_CANDIDATE_REQUIRE_UPLIFT="${WSTONE_NATIVE_RAW_CANDIDATE_REQUIRE_UPLIFT:-0}"
 NATIVE_RAW_HARDEN_TOP_GAPS="${WSTONE_NATIVE_RAW_HARDEN_TOP_GAPS:-0}"
+NATIVE_RAW_SIGNAL_TARGETED_VARIANTS="${WSTONE_NATIVE_RAW_SIGNAL_TARGETED_VARIANTS:-0}"
+NATIVE_RAW_SIGNAL_TARGETED_MAX_VARIANTS="${WSTONE_NATIVE_RAW_SIGNAL_TARGETED_MAX_VARIANTS:-6}"
 NATIVE_RAW_TOP_GAP_WEIGHTED_SELECT="${WSTONE_NATIVE_RAW_TOP_GAP_WEIGHTED_SELECT:-0}"
 NATIVE_RAW_TOP_GAP_BACKLOG_FILE="${WSTONE_NATIVE_RAW_TOP_GAP_BACKLOG_FILE:-}"
 NATIVE_RAW_TOP_GAP_REQUIRE_UPLIFT="${WSTONE_NATIVE_RAW_TOP_GAP_REQUIRE_UPLIFT:-0}"
@@ -123,6 +125,7 @@ NATIVE_SINGLESHOT_PROFILE_SHAPE_JSON='{}'
 NATIVE_MULTISHOT_JSON='{}'
 NATIVE_RAW_CANDIDATE_SEARCH_JSON='{}'
 NATIVE_RAW_HARDENING_JSON='{}'
+NATIVE_RAW_SIGNAL_TARGETED_VARIANTS_JSON='{}'
 NATIVE_RAW_TOP_GAP_REQUIREMENTS_JSON='{}'
 NATIVE_RAW_ADAPTIVE_RETRY_JSON='{}'
 INTRINSIC_REQS_JSON='[]'
@@ -429,6 +432,24 @@ if [[ "$NATIVE_RAW_CANDIDATE_SEARCH" == "1" ]]; then
    --max-variants "$NATIVE_RAW_CANDIDATE_MAX_VARIANTS" \
    --out "$OUT_DIR/02ae_raw_candidate_variants.json" >/dev/null
  VARIANTS_JSON="$(cat "$OUT_DIR/02ae_raw_candidate_variants.json")"
+  if [[ "$NATIVE_RAW_SIGNAL_TARGETED_VARIANTS" == "1" && -n "$NATIVE_RAW_TOP_GAP_BACKLOG_FILE" && -f "$NATIVE_RAW_TOP_GAP_BACKLOG_FILE" ]]; then
+    python3 "$ROOT_DIR/tools/mcp/synthesize_raw_signal_targeted_variants.py" \
+      --top-gaps "$NATIVE_RAW_TOP_GAP_BACKLOG_FILE" \
+      --max-variants "$NATIVE_RAW_SIGNAL_TARGETED_MAX_VARIANTS" \
+      --out "$OUT_DIR/02ae_raw_signal_targeted_variants.json" >/dev/null
+    signal_variants_count="$(jq '.variant_count // 0' "$OUT_DIR/02ae_raw_signal_targeted_variants.json")"
+    VARIANTS_JSON="$(jq -nc \
+      --argjson base "$VARIANTS_JSON" \
+      --argjson sig "$(cat "$OUT_DIR/02ae_raw_signal_targeted_variants.json")" \
+      '{status:"ok", active_profile_count:($base.active_profile_count // 0), variant_count:(((($base.variants // [])|length) + (($sig.variants // [])|length))), variants:(($base.variants // []) + ($sig.variants // []))}')"
+    NATIVE_RAW_SIGNAL_TARGETED_VARIANTS_JSON="$(jq -nc \
+      --argjson enabled true \
+      --arg backlog_file "$NATIVE_RAW_TOP_GAP_BACKLOG_FILE" \
+      --argjson variant_count "$signal_variants_count" \
+      '{enabled:$enabled, backlog_file:$backlog_file, variant_count:$variant_count}')"
+  else
+    NATIVE_RAW_SIGNAL_TARGETED_VARIANTS_JSON='{"enabled":false}'
+  fi

  printf '%s\n' "$TASKS" > "$OUT_DIR/02ae_candidate_0_tasks.json"
  python3 "$ROOT_DIR/tools/mcp/score_native_tasks_profile_coverage.py" \
@@ -589,6 +610,7 @@ if [[ "$NATIVE_RAW_CANDIDATE_SEARCH" == "1" ]]; then
  fi
 else
  NATIVE_RAW_CANDIDATE_SEARCH_JSON='{"enabled":false}'
+  NATIVE_RAW_SIGNAL_TARGETED_VARIANTS_JSON='{"enabled":false}'
 fi
 if [[ "$NATIVE_RAW_HARDEN_TOP_GAPS" == "1" ]]; then
  printf '%s\n' "$TASKS" > "$OUT_DIR/02af_raw_hardening_input_tasks.json"
@@ -934,6 +956,7 @@ SUMMARY_JSON="$(jq -nc \
  --argjson native_profile_autofill "$NATIVE_PROFILE_AUTOFILL_JSON" \
  --argjson native_intrinsic_boost "$NATIVE_INTRINSIC_BOOST_JSON" \
  --argjson native_raw_top_gap_requirements "$NATIVE_RAW_TOP_GAP_REQUIREMENTS_JSON" \
+  --argjson native_raw_signal_targeted_variants "$NATIVE_RAW_SIGNAL_TARGETED_VARIANTS_JSON" \
  --argjson native_raw_candidate_search "$NATIVE_RAW_CANDIDATE_SEARCH_JSON" \
  --argjson native_raw_adaptive_retry "$NATIVE_RAW_ADAPTIVE_RETRY_JSON" \
  --argjson native_raw_hardening "$NATIVE_RAW_HARDENING_JSON" \
@@ -965,6 +988,7 @@ SUMMARY_JSON="$(jq -nc \
    native_profile_autofill: $native_profile_autofill,
    native_intrinsic_boost: $native_intrinsic_boost,
    native_raw_top_gap_requirements: $native_raw_top_gap_requirements,
+    native_raw_signal_targeted_variants: $native_raw_signal_targeted_variants,
    native_raw_candidate_search: $native_raw_candidate_search,
    native_raw_adaptive_retry: $native_raw_adaptive_retry,
    native_raw_hardening: $native_raw_hardening,
--- a/tools/mcp/synthesize_raw_signal_targeted_variants.py
+++ b/tools/mcp/synthesize_raw_signal_targeted_variants.py
@@ -0,0 +1,97 @@
+#!/usr/bin/env python3
+import argparse
+import json
+from pathlib import Path
+from typing import Dict, List
+
+
+def load_json(path: Path):
+    with path.open("r", encoding="utf-8") as f:
+        return json.load(f)
+
+
+def write_json(path: Path, obj) -> None:
+    with path.open("w", encoding="utf-8") as f:
+        json.dump(obj, f, indent=2, sort_keys=True)
+        f.write("\n")
+
+
+def req_from_signal(signal: str, idx: int) -> Dict:
+    if signal.startswith("missing_prerequisite_op:"):
+        op = signal.split(":", 1)[1]
+        txt = f"raw signal-targeted policy: include prerequisite operation `{op}` in concrete tasks where applicable."
+    elif signal.startswith("missing_execution_contract:"):
+        field = signal.split(":", 1)[1]
+        txt = f"raw signal-targeted policy: set executionContract.{field}=true for generated tasks requiring safe deterministic execution."
+    elif signal.startswith("missing_reason_keyword:"):
+        key = signal.split(":", 1)[1]
+        txt = f"raw signal-targeted policy: reasons must explicitly include `{key}` risk/contract intent."
+    elif signal.startswith("native_task_count<"):
+        min_count = signal.split("<", 1)[1]
+        txt = f"raw signal-targeted policy: produce at least {min_count} concrete non-generic taskitems."
+    else:
+        txt = f"raw signal-targeted policy: address missing signal `{signal}` with concrete task structure."
+    return {
+        "requirementId": f"raw-signal-{idx+1}",
+        "kind": "constraint",
+        "normalizedText": txt,
+        "anchor": "raw_signal_targeted_variant",
+        "sourceLine": 0,
+        "ambiguous": False,
+    }
+
+
+def build_variants(prioritized_missing: List[Dict], max_variants: int) -> List[List[Dict]]:
+    if max_variants <= 0:
+        return []
+    reqs = [req_from_signal(str(row.get("missing", "")), i) for i, row in enumerate(prioritized_missing) if str(row.get("missing", ""))]
+    if not reqs:
+        return []
+
+    variants: List[List[Dict]] = []
+    # Single-signal focused variants.
+    for r in reqs:
+        variants.append([r])
+    # Pair bundles for cross-signal coupling.
+    for i in range(0, len(reqs), 2):
+        variants.append(reqs[i:i + 2])
+    # One all-signals compact bundle.
+    variants.append(reqs[: min(6, len(reqs))])
+
+    out: List[List[Dict]] = []
+    seen = set()
+    for v in variants:
+        key = tuple(x.get("requirementId", "") for x in v)
+        if not key or key in seen:
+            continue
+        seen.add(key)
+        out.append(v)
+        if len(out) >= max_variants:
+            break
+    return out
+
+
+def main() -> int:
+    p = argparse.ArgumentParser(description="Synthesize raw candidate variants targeted to top missing backlog signals.")
+    p.add_argument("--top-gaps", required=True)
+    p.add_argument("--max-variants", type=int, default=6)
+    p.add_argument("--out", required=True)
+    args = p.parse_args()
+
+    backlog = load_json(Path(args.top_gaps))
+    prioritized = list(backlog.get("prioritized_missing") or [])
+    variants = build_variants(prioritized, args.max_variants)
+
+    payload = {
+        "status": "ok",
+        "input_signal_count": len(prioritized),
+        "variant_count": len(variants),
+        "variants": variants,
+    }
+    write_json(Path(args.out), payload)
+    print(json.dumps({"status": "ok", "input_signal_count": len(prioritized), "variant_count": len(variants), "out": args.out}, sort_keys=True))
+    return 0
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())