devlyn-cli 2.0.0 → 2.2.0

package/benchmark/auto-resolve/fixtures/F27-cli-gift-card-redemption/verifiers/insufficient-balance.js

@@ -0,0 +1,36 @@
+const assert = require('node:assert');
+const fs = require('node:fs');
+const os = require('node:os');
+const path = require('node:path');
+const { spawnSync } = require('node:child_process');
+
+const workdir = process.env.BENCH_WORKDIR || process.cwd();
+const input = path.join(os.tmpdir(), `gift-card-balance-${process.pid}.json`);
+
+fs.writeFileSync(input, JSON.stringify({
+  order_id: 'order-balance',
+  lines: [
+    { sku: 'TEE', qty: 3 },
+    { sku: 'BAG', qty: 1 }
+  ],
+  redeems: [
+    { card_id: 'GC-100', amount_cents: 3000 },
+    { card_id: 'GC-100', amount_cents: 2500 }
+  ]
+}));
+
+const proc = spawnSync('node', ['bin/cli.js', 'gift-card', '--input', input], {
+  cwd: workdir,
+  encoding: 'utf8'
+});
+
+assert.strictEqual(proc.status, 2);
+assert.strictEqual(proc.stdout, '');
+assert.deepStrictEqual(JSON.parse(proc.stderr), {
+  error: 'insufficient_balance',
+  card_id: 'GC-100',
+  available_cents: 5000,
+  requested_cents: 5500
+});
+
+process.stdout.write(JSON.stringify({ ok: true }) + '\n');

package/benchmark/auto-resolve/fixtures/F27-cli-gift-card-redemption/verifiers/rules-source.js

@@ -0,0 +1,55 @@
+const assert = require('node:assert');
+const fs = require('node:fs');
+const os = require('node:os');
+const path = require('node:path');
+const { spawnSync } = require('node:child_process');
+
+const workdir = process.env.BENCH_WORKDIR || process.cwd();
+const rulesPath = path.join(workdir, 'data', 'gift-cards.json');
+const original = fs.readFileSync(rulesPath, 'utf8');
+
+try {
+  fs.writeFileSync(rulesPath, JSON.stringify({
+    products: {
+      TEE: { unit_cents: 1000 }
+    },
+    cards: {
+      'GC-SOURCE': { balance_cents: 900, active: true }
+    }
+  }, null, 2) + '\n');
+
+  const input = path.join(os.tmpdir(), `gift-card-source-${process.pid}.json`);
+  fs.writeFileSync(input, JSON.stringify({
+    order_id: 'order-source',
+    lines: [
+      { sku: 'TEE', qty: 1 }
+    ],
+    redeems: [
+      { card_id: 'GC-SOURCE', amount_cents: 700 }
+    ]
+  }));
+
+  const proc = spawnSync('node', ['bin/cli.js', 'gift-card', '--input', input], {
+    cwd: workdir,
+    encoding: 'utf8'
+  });
+
+  assert.strictEqual(proc.status, 0, proc.stderr || proc.stdout);
+  assert.strictEqual(proc.stderr, '');
+  assert.deepStrictEqual(JSON.parse(proc.stdout), {
+    order_id: 'order-source',
+    subtotal_cents: 1000,
+    gift_card_applied_cents: 700,
+    amount_due_cents: 300,
+    items: [
+      { sku: 'TEE', qty: 1, line_cents: 1000 }
+    ],
+    redemptions: [
+      { card_id: 'GC-SOURCE', applied_cents: 700, remaining_balance_cents: 200 }
+    ]
+  });
+} finally {
+  fs.writeFileSync(rulesPath, original);
+}
+
+process.stdout.write(JSON.stringify({ ok: true }) + '\n');

package/benchmark/auto-resolve/fixtures/F28-cli-rental-quote-rules/NOTES.md

@@ -0,0 +1,20 @@
+# F28 CLI rental quote rules
+
+## Why this fixture exists
+
+F27 was rejected because the direct bare arm passed every verifier. F28 returns
+to the F16 pattern that produced valid lift: exact success shape plus exact
+validation shape, with enough arithmetic and date handling that a direct
+implementation is likely to leak extra fields or miss one contract.
+
+## Pair expectation
+
+PLAN must preserve the date-counting and duplicate-combine invariants.
+IMPLEMENT must keep all public amounts in integer cents and read rules from
+`data/rental-rules.json`. VERIFY should probe both the Friday-to-Tuesday
+weekend count and the combined-stock exact error shape.
+
+## Isolation
+
+F16 covers checkout tax rules. F28 covers rental-day UTC math, weekend
+surcharges, deposits, protection fees, and non-persistent inventory validation.

package/benchmark/auto-resolve/fixtures/F28-cli-rental-quote-rules/expected.json

@@ -0,0 +1,66 @@
+{
+  "verification_commands": [
+    {
+      "cmd": "node --test tests/cli.test.js",
+      "exit_code": 0,
+      "stdout_contains": [],
+      "stdout_not_contains": ["not ok "]
+    },
+    {
+      "cmd": "node \"$BENCH_FIXTURE_DIR/verifiers/exact-success.js\"",
+      "exit_code": 0,
+      "stdout_contains": ["\"ok\":true"],
+      "stdout_not_contains": [],
+      "contract_refs": [
+        "A Friday-to-Tuesday rental counts four rental days and two weekend days.",
+        "Duplicate SKUs are combined before stock validation and pricing.",
+        "A successful quote emits exact integer-cent weekend surcharge, discount, protection, deposit, and total fields."
+      ]
+    },
+    {
+      "cmd": "node \"$BENCH_FIXTURE_DIR/verifiers/unavailable-inventory.js\"",
+      "exit_code": 0,
+      "stdout_contains": ["\"ok\":true"],
+      "stdout_not_contains": [],
+      "contract_refs": [
+        "A combined quantity over stock exits `2`, prints one JSON error to stderr, and prints no stdout.",
+        "Combined quantity over stock uses exact error shape `{ \"error\": \"unavailable_inventory\", \"sku\": string, \"available\": number, \"requested\": number }`.",
+        "The unavailable inventory error object includes `error`, `sku`, `available`, and `requested`."
+      ]
+    },
+    {
+      "cmd": "node \"$BENCH_FIXTURE_DIR/verifiers/rules-source.js\"",
+      "exit_code": 0,
+      "stdout_contains": ["\"ok\":true"],
+      "stdout_not_contains": [],
+      "contract_refs": [
+        "Item daily rates, stock, deposits, weekend surcharge percent, protection daily price, and coupons come from `data/rental-rules.json`. Do not hardcode those values in the command implementation.",
+        "Changing `data/rental-rules.json` rates, deposits, or surcharge settings changes command output without code changes."
+      ]
+    }
+  ],
+  "forbidden_patterns": [
+    {
+      "pattern": "catch\\s*\\([^)]*\\)\\s*\\{[^}]*return\\s+(null|undefined|''|\\{\\})",
+      "description": "silent catch returning fallback in rental quote path",
+      "files": ["bin/cli.js"],
+      "severity": "disqualifier"
+    },
+    {
+      "pattern": "catch\\s*\\([^)]*\\)\\s*\\{\\s*\\}",
+      "description": "empty catch block",
+      "files": ["bin/cli.js"],
+      "severity": "disqualifier"
+    },
+    {
+      "pattern": "/\\*\\s*eslint-disable",
+      "description": "eslint-disable without scoped justification",
+      "files": ["bin/cli.js"],
+      "severity": "disqualifier"
+    }
+  ],
+  "required_files": ["bin/cli.js", "tests/cli.test.js", "data/rental-rules.json"],
+  "forbidden_files": [],
+  "max_deps_added": 0,
+  "spec_output_files": ["bin/cli.js", "tests/cli.test.js"]
+}

package/benchmark/auto-resolve/fixtures/F28-cli-rental-quote-rules/metadata.json

@@ -0,0 +1,10 @@
+{
+  "id": "F28-cli-rental-quote-rules",
+  "category": "high-risk",
+  "difficulty": "high",
+  "timeout_seconds": 1500,
+  "required_tools": ["node"],
+  "browser": false,
+  "deps_change_expected": false,
+  "intent": "Add a bench-cli rental-quote command that prices equipment rentals from data/rental-rules.json using rental-day, weekend surcharge, discount, protection, deposit, and combined stock-validation rules."
+}

package/benchmark/auto-resolve/fixtures/F28-cli-rental-quote-rules/setup.sh

@@ -0,0 +1,23 @@
+#!/usr/bin/env bash
+# F28 setup - seed rental quote rules.
+set -e
+
+mkdir -p data
+
+cat > data/rental-rules.json <<'JSON'
+{
+  "items": {
+    "CAM": { "daily_cents": 1200, "stock": 2, "deposit_cents": 5000 },
+    "LIGHT": { "daily_cents": 700, "stock": 3, "deposit_cents": 2000 },
+    "TRIPOD": { "daily_cents": 400, "stock": 5, "deposit_cents": 1000 }
+  },
+  "weekend_surcharge_percent": 25,
+  "protection_daily_cents": 300,
+  "coupons": {
+    "LONG3": { "percent": 10, "min_rental_days": 3 },
+    "NONE": { "percent": 0, "min_rental_days": 1 }
+  }
+}
+JSON
+
+exit 0

package/benchmark/auto-resolve/fixtures/F28-cli-rental-quote-rules/spec.md

@@ -0,0 +1,66 @@
+---
+id: "F28-cli-rental-quote-rules"
+title: "Rental quote command with weekend and deposit rules"
+status: planned
+complexity: high
+depends-on: []
+---
+
+# F28 Rental quote command with weekend and deposit rules
+
+## Context
+
+`bench-cli` currently has greeting and version commands only. The task:
+add a `rental-quote` command that reads an equipment rental request, prices it
+from `data/rental-rules.json`, validates combined inventory, and prints one
+exact integer-cents quote.
+
+Rental quotes are operational contracts: duplicate item rows must be combined
+before stock validation, weekend surcharge rules must be deterministic, and
+success output must stay machine-readable without extra fields.
+
+## Requirements
+
+- [ ] `bench-cli rental-quote --input <path>` reads JSON shaped as `{ "start_date": "YYYY-MM-DD", "end_date": "YYYY-MM-DD", "coupon": string | null, "protection": boolean, "items": [{ "sku": string, "qty": number }] }`.
+- [ ] Item daily rates, stock, deposits, weekend surcharge percent, protection daily price, and coupons come from `data/rental-rules.json`. Do not hardcode those values in the command implementation.
+- [ ] `rental_days` is the number of calendar days from `start_date` inclusive to `end_date` exclusive, using UTC date math. `end_date` must be after `start_date`.
+- [ ] A rental day is a weekend day when its UTC day is Saturday or Sunday.
+- [ ] Combine duplicate SKUs before validating stock and before computing item rows. The output `items` array must contain one row per SKU in first-seen order.
+- [ ] Validation happens before any quote is printed. Invalid JSON, missing `items`, unknown SKU, non-positive or non-integer `qty`, combined quantity over stock, invalid date format, `end_date` not after `start_date`, unknown coupon, or non-boolean `protection` exits `2` and writes exactly one JSON error object to stderr.
+- [ ] Combined quantity over stock uses exact error shape `{ "error": "unavailable_inventory", "sku": string, "available": number, "requested": number }`.
+- [ ] `subtotal_cents` is the sum of `daily_cents * combined_qty * rental_days` for all item rows.
+- [ ] `weekend_surcharge_cents` is the sum of `Math.round(daily_cents * combined_qty * weekend_days * weekend_surcharge_percent / 100)` for all item rows.
+- [ ] `discount_cents` is `Math.round((subtotal_cents + weekend_surcharge_cents) * coupon.percent / 100)` when a coupon is present and `rental_days >= coupon.min_rental_days`; otherwise `0`.
+- [ ] `protection_cents` is `protection_daily_cents * total_combined_qty * rental_days` when `protection` is true; otherwise `0`.
+- [ ] `deposit_cents` is the sum of `deposit_cents * combined_qty` for all item rows. Deposits are never discounted.
+- [ ] `total_cents = subtotal_cents + weekend_surcharge_cents - discount_cents + protection_cents + deposit_cents`.
+- [ ] On success, write exactly one JSON object to stdout and no stderr. Keys: `rental_days`, `weekend_days`, `subtotal_cents`, `weekend_surcharge_cents`, `discount_cents`, `protection_cents`, `deposit_cents`, `total_cents`, `items`.
+- [ ] Each output item row has keys `sku`, `qty`, `rental_cents`, `deposit_cents`. `rental_cents` is `daily_cents * combined_qty * rental_days`, and row `deposit_cents` is the item's deposit times combined quantity.
+- [ ] `tests/cli.test.js` is updated. Existing tests still pass AND at least two new tests cover `rental-quote`: one successful quote and one validation failure.
+
+## Constraints
+
+- **No new npm dependencies.**
+- **No floating-money output.** All public amounts are integer cents.
+- **No silent catches.** If parsing or file reading fails, emit a visible JSON error to stderr and exit `2`.
+- **No extra stdout/stderr text** on the success path; downstream tooling parses stdout as JSON.
+- **Lifecycle note.** The harness's DOCS phase flips this spec's frontmatter `status` after implementation completes — that is benchmark lifecycle bookkeeping, not a scope violation.
+
+## Out of Scope
+
+- Persisting reservations or mutating inventory.
+- Hourly pricing, time zones beyond UTC date math, holidays, or blackout dates.
+- Taxes, shipping, currencies, or payment capture.
+- Adding server routes or web UI.
+- Touching `server/`, `web/`, or `tests/server.test.js`.
+
+## Verification
+
+- `node --test tests/cli.test.js` exits 0.
+- A Friday-to-Tuesday rental counts four rental days and two weekend days.
+- Duplicate SKUs are combined before stock validation and pricing.
+- A successful quote emits exact integer-cent weekend surcharge, discount, protection, deposit, and total fields.
+- A combined quantity over stock exits `2`, prints one JSON error to stderr, and prints no stdout.
+- The unavailable inventory error object includes `error`, `sku`, `available`, and `requested`.
+- Changing `data/rental-rules.json` rates, deposits, or surcharge settings changes command output without code changes.
+- `git diff --stat` shows only `bin/cli.js` and `tests/cli.test.js` touched (the rental rules seed comes from setup, not the arm).

package/benchmark/auto-resolve/fixtures/F28-cli-rental-quote-rules/task.txt

@@ -0,0 +1,11 @@
+Add a `rental-quote` command to `bench-cli` so users can run `bench-cli rental-quote --input <path>` with a JSON file shaped as `{ "start_date": "YYYY-MM-DD", "end_date": "YYYY-MM-DD", "coupon": string | null, "protection": boolean, "items": [{ "sku": string, "qty": number }] }`.
+
+Read item daily rates, stock, deposits, weekend surcharge percent, protection daily price, and coupons from `data/rental-rules.json`. Do not hardcode those values. `rental_days` is the number of calendar days from `start_date` inclusive to `end_date` exclusive using UTC date math. A rental day is a weekend day when its UTC day is Saturday or Sunday.
+
+Combine duplicate SKUs before stock validation and pricing. On success, write one JSON object to stdout and no stderr with keys `rental_days`, `weekend_days`, `subtotal_cents`, `weekend_surcharge_cents`, `discount_cents`, `protection_cents`, `deposit_cents`, `total_cents`, and `items`. Each item row has keys `sku`, `qty`, `rental_cents`, and `deposit_cents`.
+
+Use these formulas: `subtotal_cents` is the sum of `daily_cents * combined_qty * rental_days`; `weekend_surcharge_cents` is the sum of `Math.round(daily_cents * combined_qty * weekend_days * weekend_surcharge_percent / 100)`; `discount_cents` is `Math.round((subtotal_cents + weekend_surcharge_cents) * coupon.percent / 100)` when a coupon is present and `rental_days >= coupon.min_rental_days`, otherwise `0`; `protection_cents` is `protection_daily_cents * total_combined_qty * rental_days` when `protection` is true, otherwise `0`; `deposit_cents` is the sum of item deposits times combined quantities; `total_cents = subtotal_cents + weekend_surcharge_cents - discount_cents + protection_cents + deposit_cents`.
+
+Validation happens before any quote is printed. Invalid JSON, missing `items`, unknown SKU, non-positive or non-integer `qty`, combined quantity over stock, invalid date format, `end_date` not after `start_date`, unknown coupon, or non-boolean `protection` exits `2` and writes exactly one JSON error object to stderr. Combined quantity over stock must use exact shape `{ "error": "unavailable_inventory", "sku": string, "available": number, "requested": number }`.
+
+Update `tests/cli.test.js` so existing tests still pass and at least two new tests cover `rental-quote`: one successful quote and one validation failure.

package/benchmark/auto-resolve/fixtures/F28-cli-rental-quote-rules/verifiers/exact-success.js

@@ -0,0 +1,44 @@
+const assert = require('node:assert');
+const fs = require('node:fs');
+const os = require('node:os');
+const path = require('node:path');
+const { spawnSync } = require('node:child_process');
+
+const workdir = process.env.BENCH_WORKDIR || process.cwd();
+const input = path.join(os.tmpdir(), `rental-success-${process.pid}.json`);
+
+fs.writeFileSync(input, JSON.stringify({
+  start_date: '2026-05-08',
+  end_date: '2026-05-12',
+  coupon: 'LONG3',
+  protection: true,
+  items: [
+    { sku: 'CAM', qty: 1 },
+    { sku: 'LIGHT', qty: 1 },
+    { sku: 'CAM', qty: 1 }
+  ]
+}));
+
+const proc = spawnSync('node', ['bin/cli.js', 'rental-quote', '--input', input], {
+  cwd: workdir,
+  encoding: 'utf8'
+});
+
+assert.strictEqual(proc.status, 0, proc.stderr || proc.stdout);
+assert.strictEqual(proc.stderr, '');
+assert.deepStrictEqual(JSON.parse(proc.stdout), {
+  rental_days: 4,
+  weekend_days: 2,
+  subtotal_cents: 12400,
+  weekend_surcharge_cents: 1550,
+  discount_cents: 1395,
+  protection_cents: 3600,
+  deposit_cents: 12000,
+  total_cents: 28155,
+  items: [
+    { sku: 'CAM', qty: 2, rental_cents: 9600, deposit_cents: 10000 },
+    { sku: 'LIGHT', qty: 1, rental_cents: 2800, deposit_cents: 2000 }
+  ]
+});
+
+process.stdout.write(JSON.stringify({ ok: true }) + '\n');

package/benchmark/auto-resolve/fixtures/F28-cli-rental-quote-rules/verifiers/rules-source.js

@@ -0,0 +1,58 @@
+const assert = require('node:assert');
+const fs = require('node:fs');
+const os = require('node:os');
+const path = require('node:path');
+const { spawnSync } = require('node:child_process');
+
+const workdir = process.env.BENCH_WORKDIR || process.cwd();
+const rulesPath = path.join(workdir, 'data', 'rental-rules.json');
+const original = fs.readFileSync(rulesPath, 'utf8');
+
+try {
+  fs.writeFileSync(rulesPath, JSON.stringify({
+    items: {
+      CAM: { daily_cents: 1000, stock: 3, deposit_cents: 3000 }
+    },
+    weekend_surcharge_percent: 50,
+    protection_daily_cents: 100,
+    coupons: {
+      HALF: { percent: 50, min_rental_days: 1 }
+    }
+  }, null, 2) + '\n');
+
+  const input = path.join(os.tmpdir(), `rental-source-${process.pid}.json`);
+  fs.writeFileSync(input, JSON.stringify({
+    start_date: '2026-05-09',
+    end_date: '2026-05-11',
+    coupon: 'HALF',
+    protection: true,
+    items: [
+      { sku: 'CAM', qty: 1 }
+    ]
+  }));
+
+  const proc = spawnSync('node', ['bin/cli.js', 'rental-quote', '--input', input], {
+    cwd: workdir,
+    encoding: 'utf8'
+  });
+
+  assert.strictEqual(proc.status, 0, proc.stderr || proc.stdout);
+  assert.strictEqual(proc.stderr, '');
+  assert.deepStrictEqual(JSON.parse(proc.stdout), {
+    rental_days: 2,
+    weekend_days: 2,
+    subtotal_cents: 2000,
+    weekend_surcharge_cents: 1000,
+    discount_cents: 1500,
+    protection_cents: 200,
+    deposit_cents: 3000,
+    total_cents: 4700,
+    items: [
+      { sku: 'CAM', qty: 1, rental_cents: 2000, deposit_cents: 3000 }
+    ]
+  });
+} finally {
+  fs.writeFileSync(rulesPath, original);
+}
+
+process.stdout.write(JSON.stringify({ ok: true }) + '\n');

package/benchmark/auto-resolve/fixtures/F28-cli-rental-quote-rules/verifiers/unavailable-inventory.js

@@ -0,0 +1,35 @@
+const assert = require('node:assert');
+const fs = require('node:fs');
+const os = require('node:os');
+const path = require('node:path');
+const { spawnSync } = require('node:child_process');
+
+const workdir = process.env.BENCH_WORKDIR || process.cwd();
+const input = path.join(os.tmpdir(), `rental-stock-${process.pid}.json`);
+
+fs.writeFileSync(input, JSON.stringify({
+  start_date: '2026-05-08',
+  end_date: '2026-05-10',
+  coupon: null,
+  protection: false,
+  items: [
+    { sku: 'CAM', qty: 1 },
+    { sku: 'CAM', qty: 2 }
+  ]
+}));
+
+const proc = spawnSync('node', ['bin/cli.js', 'rental-quote', '--input', input], {
+  cwd: workdir,
+  encoding: 'utf8'
+});
+
+assert.strictEqual(proc.status, 2);
+assert.strictEqual(proc.stdout, '');
+assert.deepStrictEqual(JSON.parse(proc.stderr), {
+  error: 'unavailable_inventory',
+  sku: 'CAM',
+  available: 2,
+  requested: 3
+});
+
+process.stdout.write(JSON.stringify({ ok: true }) + '\n');

package/benchmark/auto-resolve/fixtures/SCHEMA.md

@@ -73,7 +73,8 @@ Machine-readable acceptance criteria used by both `run-fixture.sh` verification
       "cmd": "node bin/cli.js doctor",
       "exit_code": 0,
       "stdout_contains": ["doctor: "],
-      "stdout_not_contains": ["undefined"]
+      "stdout_not_contains": ["undefined"],
+      "contract_refs": []
     }
   ],
   "forbidden_patterns": [
@@ -86,14 +87,25 @@ Machine-readable acceptance criteria used by both `run-fixture.sh` verification
   ],
   "required_files": ["bin/cli.js"],
   "forbidden_files": [],
+  "tier_a_waivers": [],
+  "spec_output_files": ["bin/cli.js"],
   "max_deps_added": 0
 }
 ```
 
 - **verification_commands** — runner executes each. Each command's pass/fail contributes to the arm's `verify_score`.
+  Commands run with `BENCH_WORKDIR` (fresh arm work tree) and
+  `BENCH_FIXTURE_DIR` (the fixture directory outside the arm work tree) in
+  the environment. Put discriminator/oracle scripts under the fixture
+  directory when the arm should not read the verifier source.
+  Any command that references `BENCH_FIXTURE_DIR` is a hidden oracle and must
+  include `contract_refs`: exact substrings from `spec.md` proving the oracle
+  tests a visible contract rather than inventing a narrower one.
 - **forbidden_patterns** — regexes scanned across `diff.patch`. Match at `severity: "disqualifier"` is a hard-floor fail. Match at `severity: "warning"` goes into the judge's critical-findings report.
 - **required_files** — must exist after the arm runs.
 - **forbidden_files** — must NOT appear in the arm's diff.
+- **tier_a_waivers** — optional globs for files the spec explicitly authorizes even though Tier A scope oracle would normally flag them.
+- **spec_output_files** — files or globs that define the authorized output surface for Tier B scope tracing.
 - **max_deps_added** — count of new entries under `dependencies`/`devDependencies` in `package.json`. Exceeds → hard-floor fail.
 
 ## NOTES.md

package/benchmark/auto-resolve/scripts/collect-swebench-predictions.py

@@ -0,0 +1,98 @@
+#!/usr/bin/env python3
+"""Collect patch.diff files into SWE-bench prediction JSONL."""
+
+from __future__ import annotations
+
+import argparse
+import json
+from pathlib import Path
+from typing import Any
+
+
+def read_jsonl(path: Path) -> list[dict[str, Any]]:
+    rows: list[dict[str, Any]] = []
+    with path.open(encoding="utf8") as f:
+        for line_no, line in enumerate(f, start=1):
+            if not line.strip():
+                continue
+            value = json.loads(line)
+            if not isinstance(value, dict):
+                raise ValueError(f"{path}:{line_no}: expected JSON object")
+            rows.append(value)
+    return rows
+
+
+def instance_ids_from_jsonl(path: Path | None) -> set[str] | None:
+    if path is None:
+        return None
+    ids: set[str] = set()
+    for row in read_jsonl(path):
+        instance_id = row.get("instance_id")
+        if not isinstance(instance_id, str) or not instance_id:
+            raise ValueError(f"{path}: row missing non-empty instance_id")
+        ids.add(instance_id)
+    return ids
+
+
+def collect_from_root(root: Path, patch_name: str, keep: set[str] | None) -> list[tuple[str, Path]]:
+    patches: list[tuple[str, Path]] = []
+    for patch_path in sorted(root.glob(f"*/{patch_name}")):
+        instance_id = patch_path.parent.name
+        if keep is not None and instance_id not in keep:
+            continue
+        patches.append((instance_id, patch_path))
+    return patches
+
+
+def main() -> int:
+    parser = argparse.ArgumentParser()
+    parser.add_argument("--patch-root", required=True, type=Path)
+    parser.add_argument("--patch-name", default="patch.diff")
+    parser.add_argument("--instances-jsonl", type=Path, help="Optional filter/validation set.")
+    parser.add_argument("--model-name", required=True)
+    parser.add_argument("--out", required=True, type=Path)
+    parser.add_argument("--allow-empty", action="store_true")
+    args = parser.parse_args()
+
+    keep = instance_ids_from_jsonl(args.instances_jsonl)
+    patches = collect_from_root(args.patch_root, args.patch_name, keep)
+    if not patches:
+        raise ValueError(f"no {args.patch_name} files found under {args.patch_root}")
+
+    args.out.parent.mkdir(parents=True, exist_ok=True)
+    written = 0
+    skipped_empty: list[str] = []
+    with args.out.open("w", encoding="utf8") as f:
+        for instance_id, patch_path in patches:
+            patch = patch_path.read_text(encoding="utf8")
+            if not patch.strip():
+                if args.allow_empty:
+                    skipped_empty.append(instance_id)
+                    continue
+                raise ValueError(f"empty patch for {instance_id}: {patch_path}")
+            f.write(
+                json.dumps(
+                    {
+                        "instance_id": instance_id,
+                        "model_name_or_path": args.model_name,
+                        "model_patch": patch,
+                    }
+                )
+                + "\n"
+            )
+            written += 1
+
+    report = {
+        "patch_root": str(args.patch_root),
+        "patch_name": args.patch_name,
+        "model_name_or_path": args.model_name,
+        "out": str(args.out),
+        "predictions_written": written,
+        "empty_skipped": skipped_empty,
+    }
+    print(json.dumps(report, indent=2))
+    return 0
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())