Shape broad mathematical interests into precise research questions. Search prior work on definitions and theorems, record traceable sources, map supporting and conflicting evidence, and identify gaps that could lead to testable conjectures.
数学问题发现与证据检索。用于界定研究问题、查定义/定理谱系、检索 arXiv/Scholar/OpenAlex/Crossref、建立来源账本、证据图、查新或从证据缺口生成可证伪猜想。
Review the instructions the agent will receive before you install.
把模糊兴趣变成可界定、可检索、可证伪的数学问题，并产出来源可追溯的证据图。
本 skill 负责方法地图第一层的“规格与语义”前置：在选择 Lean、SMT、model checking 或 abstract interpretation 之前，先冻结对象、定义域、量词、来源和可接受结论。对应的机器边界是 ProblemContract；地图总览见 FORMAL-METHODS-MAP.md。
research_eligible=false；不得直接创建 Attempt，也不得把来源的 answered/resolved/solved 当作数学 Result。--collection candidates|all，并在输出中保留 collection、来源和许可边界。1. 固定对象、领域、问题和非目标。
2. 先查 `query_problem_library.py --collection admitted`；需要扩面时再显式查 candidates。
3. 建立术语：正式名、别名、旧名、符号、MSC、相邻领域术语。
4. 冻结检索式、来源、日期、语言和停止条件。
5. 优先原始论文、正式出版物、arXiv 原文和官方数据库记录。
6. 为每个来源记录稳定 ID、URL/DOI/arXiv ID、版本、raw locator 和证据位置。
7. 将关系标为 supports / contradicts / limits / extends / unknown；候选身份匹配只进入 review queue。
8. 输出已知事实、冲突、空白、候选猜想和下一步取证。
默认 provider 顺序：项目资源/MCP → SearXNG arxiv,semantic scholar,openalex,crossref →通用 Web。429/CAPTCHA 时记录失败并切换 provider；不无限重试。
research_eligible=false，唯一下一步是来源/陈述准入或 ProblemContract 冻结。references/source-map.md：研究方法与检索供应链映射。references/pressure-tests.md：查新与摘要误用压力场景。rw-research-skill、wentor-research-plugins、kdense-scientific-skills。No reviews yet. Tell others what you thought.
No login needed. Share how it worked for you, ask a question, or note where it did not run.
No comments yet. Be the first to write one.
Analyze Solidity smart contracts with Slither and Mythril before Ethereum deployment. Combine code inspection with execution path analysis, review potential vulnerabilities, remove duplicate findings, and produce an audit report with severity ratings and suggested fixes.
Audit Solidity contracts in Foundry projects before deployment using code analysis, randomized tests, manual review, and private key checks. Produces a JSON report with findings, remediation, test coverage, and a pass or fail deployment decision.
Build reusable Claude Skills from documentation, APIs, code, and specifications. Organize instructions and reference files, define clear activation conditions, add reproducible examples, and check new or existing skills against a delivery checklist.
Automate tmux, a tool for managing multiple terminal workspaces, to inspect output, send commands, monitor stalled tasks, and record logs. Coordinate multiple AI terminals with task queues, file locks, status tracking, and handoff reports.
Check mathematical expressions, solve equations, and search bounded ranges for counterexamples using symbolic and numerical libraries. Record assumptions, commands, and results so calculations can be repeated, while keeping computational evidence distinct from general mathematical proof.
Turn scattered formulas and theory notes into a derivation that can be checked step by step. Keep definitions and assumptions consistent, distinguish exact identities from approximations, and make hidden conditions and limits explicit.