{"data":{"kind":"file","path":"README.md","version_id":"cehv6nklw4fr7fths9lheo7q","entry":{"name":"README.md","path":"README.md","is_directory":false,"size":2256,"modified_at":"2026-08-27T06:58:07.523000","content_hash":"c4d84b69cecefc6016eb4cc7b071bd22c6018653d44f5b5559011ed1b54c80e2"},"entries":[],"content":"# Reverse Engineering / Decompiler Challenge\n\nA verifiers environment for training and evaluating LLMs on reverse engineering and decompilation tasks. Models must analyze low-level, obfuscated, or compiled code and reconstruct readable high-level source code with accurate logic identification.\n\n## Challenge Types\n\n| Type | Description | Difficulty |\n|------|-------------|------------|\n| **Assembly** | x86/x64 assembly to C decompilation | 1-3 |\n| **Bytecode** | Python bytecode disassembly analysis | 2-3 |\n| **Obfuscated** | Deobfuscation of obfuscated JS/Python | 2-3 |\n| **Regex** | Compiled regex pattern recovery | 1-2 |\n| **Control Flow** | CFG to high-level code reconstruction | 2-3 |\n| **Crypto** | Cryptographic algorithm identification | 1-3 |\n| **Symbol Recovery** | Stripped binary function identification | 2-3 |\n\n## Scoring Dimensions\n\nThe environment evaluates responses on four dimensions:\n\n1. **Accuracy (35%)** - Correct algorithm/function identification via keyword matching\n2. **Code Quality (25%)** - Structure, readability, and completeness of reconstructed code\n3. **Completeness (20%)** - Coverage of all requested analysis points\n4. **Technical Depth (20%)** - Domain-specific knowledge demonstration\n\n## Usage\n\n```python\nimport verifiers as vf\n\nenv = vf.load_environment(\"re-decompiler-challenge\")\n# or with difficulty filter:\nenv = vf.load_environment(\"re-decompiler-challenge\", difficulty=2)\n```\n\n## Configuration\n\n| Parameter | Type | Default | Description |\n|-----------|------|---------|-------------|\n| `difficulty` | int | None (all) | Max difficulty level to include (1-3) |\n\n## Example Challenges\n\n- Decompile factorial/binary-search/merge-sort from x86-64 assembly\n- Reconstruct Python source from bytecode disassembly\n- Deobfuscate XOR cipher and lambda calculus expressions\n- Identify RC4 and Base64 implementations from decompiled C\n- Recover function names from stripped binary code\n\n## Research Applications\n\nThis environment is useful for:\n- Evaluating code understanding beyond surface-level pattern matching\n- Testing low-level systems knowledge (calling conventions, register allocation)\n- Measuring ability to reason about compiled/transformed code\n- Security analysis and vulnerability research training\n","encoding":"utf-8","truncated":false,"total_bytes":2256},"status":null}