Adds two direct-placement BBOPlace Frontier-CS 2.0 tasks

2.0/README.md

@@ -52,3 +52,20 @@ BBOPlace-Bench. Its problem ID is `bboplace_iccad2015`. It follows the same
 candidate format, CPU-only evaluator, MP-HPWL metric, relaxed MGO baselines,
 and quick-versus-final evaluation flow as `bboplace_ispd2005`, but scores the
 ICCAD2015 benchmark set.
+
+## BBOPlace Direct ISPD2005
+
+This direct-placement variant asks agents to submit one JSON placement for a
+single ISPD2005 design, `adaptec1`, instead of writing a Python placement
+generator. Its problem ID is `bboplace_direct_ispd2005`. The evaluator uses the
+same CPU-only BBOPlace MGO MP-HPWL path and relaxed baseline as the ISPD2005
+suite task, but both iterative feedback and final verification score only that
+one design.
+
+## BBOPlace Direct ICCAD2015
+
+This direct-placement variant asks agents to submit one JSON placement for a
+single ICCAD2015 design, `superblue1`. Its problem ID is
+`bboplace_direct_iccad2015`. It follows the same JSON interface and single
+design evaluation flow as `bboplace_direct_ispd2005`, with the ICCAD2015
+baseline for `superblue1`.

2.0/problems/bboplace_direct_iccad2015/config.yaml

@@ -0,0 +1,18 @@
+tag: optimization
+runtime:
+  language: json
+  timeout_seconds: 10800
+  environment: "JSON placement for one hidden ICCAD2015 BBOPlace design"
+  apt_packages:
+    - python3-numpy
+  docker:
+    image: ubuntu:24.04
+    judge_image: ghcr.io/frontiercs/frontiercs-bboplace-data:2026-06-ispd-iccad
+submission:
+  kind: file
+  path: /app/solution.json
+environment:
+  cpus: 8
+  memory_mb: 16384
+  storage_mb: 8192
+  build_timeout_seconds: 3600

2.0/problems/bboplace_direct_iccad2015/evaluate.sh

@@ -0,0 +1,12 @@
+#!/usr/bin/env bash
+set -euo pipefail
+
+SCRIPT_DIR=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)
+SOLUTION="/work/execution_env/solution_env/solution.json"
+
+if [[ ! -f "$SOLUTION" ]]; then
+  echo "Error: Missing $SOLUTION" >&2
+  exit 1
+fi
+
+python "$SCRIPT_DIR/evaluator.py" "$SOLUTION"