{"product_id":"ai-agent-skill-reward-hackability-auditor-for-rl-environments","title":"AI Agent Skill: Reward Hackability Auditor for RL Environments","description":"\u003ch3\u003eAI Agent Skill: Reward Hackability Auditor for RL Environments\u003c\/h3\u003e\n\n\u003cp\u003eThis skill audits reinforcement learning environment verifiers, reward functions, and grading rubrics for reward-hacking vulnerabilities before training or publishing. It is designed for use in inspecting, designing, testing or strengthening RL environments across multiple formats.\u003c\/p\u003e\n\n\u003ch3\u003eWhat this skill does\u003c\/h3\u003e\n\u003cul\u003e\n    \u003cli\u003eExamines RL environments for reward-hacking exploitability across six research-backed exploit classes.\u003c\/li\u003e\n    \u003cli\u003ePrevents agents from exploiting flawed grading logic rather than solving tasks legitimately.\u003c\/li\u003e\n    \u003cli\u003eAnalyzes reward functions and verifiers in OpenEnv, Prime Intellect verifiers-spec, and Gymnasium formats.\u003c\/li\u003e\n    \u003cli\u003eUtilizes native schema detection and abstract syntax tree (AST) parsing for comprehensive auditing of code elements.\u003c\/li\u003e\n\u003c\/ul\u003e\n\n\u003ch3\u003eWho it is for\u003c\/h3\u003e\n\u003cp\u003eThis skill is intended for developers and teams using AI coding agents such as Claude Code, Cursor, and Codex. It is especially beneficial for those involved in developing, testing, or auditing reinforcement learning environments.\u003c\/p\u003e\n\n\u003ch3\u003eUse cases\u003c\/h3\u003e\n\u003cul\u003e\n    \u003cli\u003eInspecting RL environments before training to ensure robustness against reward manipulation.\u003c\/li\u003e\n    \u003cli\u003eDesigning new environments with security and integrity in mind by guarding against prevalent exploit classes.\u003c\/li\u003e\n    \u003cli\u003eTesting existing environments to identify and remedy vulnerabilities in reward functions and grading rubrics.\u003c\/li\u003e\n    \u003cli\u003eHardening published RL environments by verifying their verifier and reward function against known vulnerabilities.\u003c\/li\u003e\n\u003c\/ul\u003e\n\n\u003ch3\u003eTechnical details\u003c\/h3\u003e\n\u003cul\u003e\n    \u003cli\u003eSupports OpenEnv format using key files like \u003ccode\u003eopenenv.yaml\u003c\/code\u003e, \u003ccode\u003eserver\/app.py\u003c\/code\u003e, and others through native schema detection \u0026amp; AST parsing.\u003c\/li\u003e\n    \u003cli\u003eCompatible with Prime Intellect \u003ccode\u003everifiers-spec\u003c\/code\u003e by analyzing entry points and class hierarchies, including \u003ccode\u003eload_environment()\u003c\/code\u003e and associated markers.\u003c\/li\u003e\n    \u003cli\u003eEvaluates Gymnasium environments by extracting step rewards and conducting AST analysis on methods like \u003ccode\u003estep()\u003c\/code\u003e and \u003ccode\u003ereset()\u003c\/code\u003e.\u003c\/li\u003e\n    \u003cli\u003eOffers fallback heuristic scanning for raw Python setups using files such as \u003ccode\u003everifier.py\u003c\/code\u003e and \u003ccode\u003ereward.py\u003c\/code\u003e.\u003c\/li\u003e\n\u003c\/ul\u003e\n\n\u003cp\u003eFor authorized security testing, defensive research, and educational use only.\u003c\/p\u003e\n\u003c!-- mcpcart:static-blocks:start --\u003e\n\u003chr\u003e\n\u003ch3\u003eSource \u0026amp; Licence\u003c\/h3\u003e\n\u003cp\u003eThis package is built on open-source work published by \u003cstrong\u003eFreakyAdy\u003c\/strong\u003e (\u003ca href=\"https:\/\/github.com\/FreakyAdy\/Reward-Hackability-Auditor--CLI---Claude-Skill-\" rel=\"nofollow noopener\" target=\"_blank\"\u003eFreakyAdy\/Reward-Hackability-Auditor--CLI---Claude-Skill-\u003c\/a\u003e) and distributed under \u003cstrong\u003eMIT\u003c\/strong\u003e. The original licence text and copyright notice are included in your download.\u003c\/p\u003e\n\u003cp\u003ePersonal and commercial use, modification and redistribution are permitted, provided the original copyright and licence notice are retained.\u003c\/p\u003e\n\u003cp\u003eYour purchase covers curation, licence verification, packaging, documentation and instant delivery. It does not grant exclusive rights to the underlying open-source code, which remains available under its original licence.\u003c\/p\u003e\n\u003ch3\u003eDelivery \u0026amp; Support\u003c\/h3\u003e\n\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eDelivery:\u003c\/strong\u003e instant — a secure download link is emailed to you as soon as payment is confirmed.\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eFormat:\u003c\/strong\u003e ZIP archive containing the skill files, documentation and the original licence.\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eSupport:\u003c\/strong\u003e \u003ca href=\"mailto:support@mcpcart.com\"\u003esupport@mcpcart.com\u003c\/a\u003e — we aim to reply within 2 business days.\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eUpdates:\u003c\/strong\u003e updates are included only where stated on this page.\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003ch3\u003eRefunds\u003c\/h3\u003e\n\u003cp\u003eThis is a digital product delivered immediately after purchase. By completing your order you request immediate delivery and acknowledge that, once the download has been accessed, the statutory right to cancel no longer applies to the extent permitted by law. Refund requests are handled in accordance with our published Refund Policy.\u003c\/p\u003e\n\u003cp style=\"font-size:0.85em;color:#666;\"\u003eClaude, Codex, Gemini and Cursor are trademarks of their respective owners. MCP Cart is an independent marketplace and is not affiliated with, endorsed by, or sponsored by any of them. Compatibility references describe interoperability only.\u003c\/p\u003e\n\u003c!-- mcpcart:static-blocks:end --\u003e","brand":"MCP Cart","offers":[{"title":"Default Title","offer_id":52999732134199,"sku":"MCP-FREAKYADY-REWARD-HACKABILITY-AUDITOR-CLI-CLAUDE-SKILL-REWARD-HACKABILITY-AUDITOR","price":59.99,"currency_code":"GBP","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/0981\/3950\/4951\/files\/sMhmx2akOZzS3K4PbEua9_b8ac0f9eb0324f76a42997fa9d260f1b.jpg?v=1789639544","url":"https:\/\/mcpcart.com\/products\/ai-agent-skill-reward-hackability-auditor-for-rl-environments","provider":"SPF PRO","version":"1.0","type":"link"}