MiniMax AI

@minimax-ai source repo

171 skills
0 followers
13k repo stars
2 weeks ago last updated

▌
Mmx CLI 2 · minimax-ai bundle
Use mmx to generate text, images, video, and speech, and to transcribe audio, via the MiniMax AI platform. Use when the user wants to create media content, chat with MiniMax models, transcribe audio to text, perform web search, or manage MiniMax API resources from the terminal.

12.9k repo stars
▌
H3 Prompt Writing 2 · minimax-ai bundle
Write MiniMax H3 video generation prompts for T2VA, I2VA, FL2VA, L2VA, and Ref2VA. Use when rewriting multimodal requests into H3 prompt structures, composing integrated_multimodal_description, overall_soundscape, and non_diegetic_music, aligning keyframes, or defining reference labels for images, videos, and audio.

12.9k repo stars
▌
Customize 2 · minimax-ai
当用户说"修改画像/改一下我的设置/更新我的信息/换个身份/修改红线/改输出偏好/ 调整 guardrails/改规则/换数据源配置/我刚考了律师证/我换行业了"时使用。这是修改 docs/guardrails.md 执业画像与 Shared guardrails 的唯一入口：读取现状 → 与使用者确认修改内容 → 写回 → 提示本次修改影响哪些技能行为。防止直接手改跳过影响评估，也防止多人协作时画像被无声覆盖。

12.9k repo stars
▌
Task Tracker 2 · minimax-ai bundle
Use when: 任务追踪与日报周报生成。用于记录老板工作进度、生成日报周报、持续追踪任务完成情况。.

12.9k repo stars
▌
Cold Start Interview 2 · minimax-ai
当用户首次使用本插件、docs/guardrails.md 执业画像中仍存在 [填空]、用户说"初始化设置/开始访谈/ 设置我的信息/配置画像/重新做一遍设置/检查数据源配置是否生效/检查北大法宝 token"时使用。通过结构化访谈收集使用者的角色、法域、行业、常做业务、红线条款、升级矩阵、签署流程、数据源与输出偏好，并以散文形式写回 docs/guardrails.md 执业画像章节。支持 quick （3 分钟 6 问）、full（20 问）、--redo（重谈）、--check-integrations（只检查凭证）四种模式，支持中断后凭 [PENDING] 标记续谈。

12.9k repo stars
▌
Acp Task Dispatch · minimax-ai
Dispatch a self-contained task to the OpenClaw-mcode-ACP HTTP server from inside MiniMax Code. Use when a task should be persisted, retried, observed over time, or processed by a worker pool instead of the current MiniMax Code session.

12.9k repo stars
▌
Document Skills PDF · minimax-ai bundle
Create, extract, transform, or analyze PDF documents. Use this Skill whenever a .pdf file is the input or output - generating reports, one-pagers, posters or structured documents as real selectable-text PDFs; extracting text, tables or images from existing PDFs; merging, splitting, rotating, or watermarking pages; filling forms; checking that a produced PDF has the right page count and extractable text.

12.9k repo stars
▌
Labor Contract Review · minimax-ai
当用户提供劳动合同、聘用协议、录用通知（offer）或续签/变更协议文本，要求审查条款合法性、完备性或对本方利弊时使用。覆盖场景：新签劳动合同审查、续签文本比对、offer 与正式合同一致性核对、入职文件包中的合同文本审查。同义场景词：劳动合同审查、offer 审查、入职合同审查、续签合同审查。执行链路：画像与 blocks 红线检查、雇主/员工视角确认、必备条款核对、试用期合法性、服务期与违约金限定、竞业限制三要件审查、按视角三色分桶输出 memo，收尾过引用审计并提示下一步。

12.9k repo stars
▌
Privacy Policy Review · minimax-ai
当用户需要审查、修订或上架前核对隐私政策、个人信息保护政策、App 隐私条款时使用。覆盖场景：隐私政策起草后自检、监管通报或检测问题后的整改复核、版本更新合规审查、App 上架前隐私合规检查、第三方 SDK 清单与权限调用对应核查。同义场景词：隐私政策审查、隐私条款审查、隐私声明、privacy policy review。执行全链路：八大检查类别逐项过堂、🟢🟡🔴 三色分桶、输出带 reviewer note 的审查 memo，问题项附整改建议与优先级；需要重写的条款不代拟，一律建议转法务或律师起草。

12.9k repo stars
▌
Sales Contract Review · minimax-ai
买卖合同专项审查技能，由 contract-review 路由器按合同类型路由加载，不直接面向用户调用。覆盖场景：买卖合同、采购合同、购销合同、销售合同、供货协议、订单+主协议结构的货物交易文本。同义场景词：买卖合同审查、采购合同审查、购销合同审查、销售合同审查、供货协议审查。执行与 nda-review 同构的全链路： matter 上下文与产物去向检查、买方/卖方立场判定、买卖立场 playbook 加载或补齐、八类分类检查（标的物与质量、交付与风险转移、价款与支付、验收、违约责任、所有权保留、解除、管辖与争议解决）、按 playbook 三色分桶、输出统一 triage memo。

12.9k repo stars
▌
Trademark Search Prep · minimax-ai
当用户准备申请商标、注册品牌标识，需要评估标志可注册性、确定申请类别、准备申请材料时使用。覆盖场景：新品牌商标申请前评估、显著性初筛、尼斯分类选择、近似检索策略制定、申请材料清单准备。同义场景词：商标注册、商标申请准备、商标检索、LOGO 注册、品牌保护。执行链路：显著性与禁用条款初筛、核心类+防御类类别规划、官方检索指引（人工执行）、主体与材料清单、时间线与费用构成说明，收尾过引用审计并走后果门。本技能不做「保证注册成功」的结论。

12.9k repo stars
▌
Verification Loop · minimax-ai
Enforce a claim-to-proof workflow before declaring work complete. Use this Skill when a task involves implementation, debugging, review, or any other work that could be claimed complete without concrete evidence.

12.9k repo stars
▌
Background Task · minimax-ai
Background Task

12.9k repo stars
▌
Parallel Fanout · minimax-ai
Decompose a task into 2+ truly independent sub-tasks and dispatch them concurrently via separate `task()` calls. Pick whether to fan out explicitly, not by accident. USE WHEN: the user task is clearly decomposable into 2+ independent sub-tasks (independent files, independent probes, independent analyses), you would otherwise serialize work that has no real dependency, user said "in parallel" / "并行" / "fan out" / "spawn agents" / "同时跑". TRIGGER PHRASES: "in parallel", "parallel", "fan out", "spawn agents", "并行", "同时", "concurrent", "subagents", "multi-agent", "同时跑几个". SKIP WHEN: sub-tasks have a hard data dependency (output of A is input of B), the user explicitly said "sequential" / "one at a time", there is only one sub-task.

12.9k repo stars
▌
Session Handoff · minimax-ai
At session end, write a structured handoff file so next session can pick up in 30 seconds. USE WHEN: user says "今天先到这" / "done for today" / "see you tomorrow" / "we'll continue later" / "下次再继续" / "end session" / "收尾", context about to compact, long task in progress, natural pause approaching (end of work day, end of milestone), sub-task in flight that outlives this session. TRIGGER PHRASES: "今天先到这", "done for today", "see you tomorrow", "we'll continue later", "下次再继续", "先到这", "end session", "session 结束", "收尾", "写到 handoff file", "wrap up", "session handoff", "session 接力". SKIP WHEN: session just started (no in-progress work to hand off), work is fully complete and verified (completion-audit passed), user said "throw it all away, start fresh next time" / "全部扔掉".

12.9k repo stars
▌
Document Skills DOCX · minimax-ai bundle
Create, read, edit, or review Microsoft Word .docx documents. Use this Skill whenever the task involves a .docx file - generating reports, letters, or structured documents from scratch; extracting or reorganizing content; applying consistent formatting (headings, tables, styles, headers/footers, page numbers); editing an existing document while preserving its layout; or checking that a produced document actually opens and renders.

12.9k repo stars
▌
Document Skills PPTX · minimax-ai bundle
Create, read, edit, or analyze PowerPoint .pptx presentations. Use this Skill whenever the task involves slides or a deck - building a presentation from an outline or content file; editing text, tables, or charts in an existing deck; extracting slide content for review or repurposing; checking that a produced deck opens, has the right slide count, and nothing overflows its slide.

12.9k repo stars
▌
Document Skills XLSX · minimax-ai bundle
XLSX workbench

12.9k repo stars
▌
Company Charter Review · minimax-ai
当用户需要审查、修订或起草核对公司章程时使用。覆盖场景：新设公司章程审查、存量公司对照 2023 年修订公司法的章程修订、五年实缴出资安排落地、治理结构调整（审计委员会替代监事会、简化董事会等）、章程与股东协议冲突核查。同义场景词：章程审查、公司章程修订、articles of association review。执行全链路：法定必备事项核对、2023 公司法要点比对、自治空间条款（表决权差异/分红约定/股权转让限制）合法性边界检查、🟢🟡🔴 三色分桶 memo；立场（公司/控股股东/中小股东）先定，不硬编码实体立场。

12.9k repo stars
▌
Data Export Assessment · minimax-ai
当用户涉及向境外提供个人信息或重要数据、需要判断数据出境合规路径时使用。覆盖场景：数据出境安全评估申报、个人信息出境标准合同（SCC）备案、个人信息保护认证三路径选择，跨境业务上线、境外总部数据回传、跨国集团人力资源与客户数据共享、出海产品的数据架构合规评估。同义场景词：数据出境、跨境传输、 SCC 备案、出境评估、跨境数据流动。执行全链路：出境场景识别问卷、三路径判断树、材料清单、时间表与整改建议；门槛与阈值一律经 statute-verify 核验，不作「保证通过」的结论。

12.9k repo stars
▌
Data Incident Response · minimax-ai
当用户发现或怀疑发生个人信息泄露、篡改、丢失等数据安全事件，需要应急响应支持时使用。覆盖场景：数据泄露初判与分级、法定通知义务判断（补救措施、通知监管与通知个人的触发条件）、证据保全、首小时行动清单、监管版/个人版/ 内部版三套通知文本准备、事后复盘整改。同义场景词：数据泄露、拖库、安全事件应急、个保事件、数据事件响应、incident response。红线纪律：不得建议隐瞒、淡化或迟报事件；应急场景跳过画像阻塞式检查，画像缺项并行补齐。

12.9k repo stars
▌
Labor Arbitration Prep · minimax-ai
当用户准备申请劳动仲裁或收到仲裁通知需要应诉时使用。覆盖场景：劳动者主张工资、加班费、经济补偿、赔偿金、未签合同二倍工资；用人单位收到仲裁申请后的应诉材料准备；仲裁时效核查。同义场景词：劳动仲裁、仲裁申请书、仲裁证据清单、仲裁时效、仲裁管辖。执行链路：时效初筛（临近立即标红）、仲裁请求清单设计、证据清单整理、管辖确认、输出材料清单与时间表，收尾过引用审计并走后果门。

12.9k repo stars
▌
Termination Assessment · minimax-ai
当用户描述拟解除或辞退员工、协商解除谈判、经济性裁员方案，或作为劳动者评估单位解除行为合法性时使用。覆盖场景：单方解除合规评估、协商解除方案审查、裁员路径选择、离职谈判底线测算、被迫解除（单位过错）主张评估。同义场景词：辞退评估、解除合规、裁员方案、N+1 计算、违法解除判断。执行链路：事实采集与视角确认、解除路径判断树、经济补偿 N/N+1/2N 计算、程序要件核对、禁止解除情形排查、风险分级输出行动清单，收尾过引用审计。

12.9k repo stars
▌
Completion Audit · minimax-ai
Before saying "done", derive requirements, find authoritative evidence, verify each is ✅. USE WHEN: about to say "done" / "complete" / "ship it" / "I finished" / "做完了" on non-trivial task, about to mark `todowrite` step done, about to update goal to `complete`, user has been waiting for "done" for several turns, "looks good" / "should be fine" / "应该好了" / "我试过没报错" / "我跑了测试都过了" / "I tested it" / "trust me" / "should work". TRIGGER PHRASES: "做完了", "done", "complete", "ship it", "好了", "完成", "搞定", "I think we're done", "应该好了", "looks good", "我试过没报错", "我跑了测试都过了", "I tested it", "trust me", "should work". SKIP WHEN: one-line edit, user can see result in chat immediately, user explicitly said "ship it" / "no more review" in this turn.

12.9k repo stars
▌
Goal Persistence · minimax-ai
Maintain explicit north-star goal for the whole thread that survives compactions and detects drift. USE WHEN: non-trivial task stated, user redirected mid-task ("actually do X instead" / "wait scrap that" / "现在改成"), before `context-pressure-compact`, about to mark done, user said "我们的目标是" / "we're trying to" / "我想要的" / "what I want is" / "目标是", agent drifting (tool call no longer serves original ask). TRIGGER PHRASES: "我们的目标", "目标是", "我想要", "we're trying to", "what I want is", "drift", "走偏了", "focus on", "stay focused", "on track", "actually do X instead", "wait scrap that", "现在改成". SKIP WHEN: trivial one-shot task, exploration without commitment, goal hasn't changed in many turns.

12.9k repo stars
▌
Long Term Memory · minimax-ai
Design a cross-session long-term memory system that extracts, consolidates, and cites durable knowledge from conversation rollouts. USE WHEN: building any system that needs to persist insights across sessions, designing "what should the next agent remember" pipelines, building memory workspaces with git baseline diffing, planning Phase 1/Phase 2 memory architectures, writing JSON-schema-constrained extraction prompts, deciding what NOT to write (no-op gate), or any task involving "memories that survive session boundaries". TRIGGER PHRASES: "long-term memory", "cross-session memory", "memory pipeline", "memory consolidation", "memory citation", "raw_memories.md", "MEMORY.md", "phase 1 extraction", "phase 2 consolidation", "watermark", "no-op gate", "git baseline diff". SKIP WHEN: single-session task state (use `world-state-tracking` instead), ephemeral/short task, no need to survive session boundaries, in-memory only.

12.9k repo stars
▌
Plan Stream Emit · minimax-ai
Before touching files on a non-trivial task, emit a structured plan and surface to the user for early course-correction. USE WHEN: non-trivial task, multi-step task, ambiguous requirement, would take > 3 tool calls, user has not approved an approach yet, user said "plan first" / "before you start" / "let me see your approach" / "先出计划" / "出方案", crossing trust boundary (production, public repo, irreversible action). TRIGGER PHRASES: "plan first", "先出计划", "let me see", "出方案", "确认一下", "先别动手", "想清楚再开始", "before you start", "我看看方案", "出 plan", "出计划". SKIP WHEN: single one-shot question, user already gave numbered list of steps, trivially reversible, "do X" with X being one line.

12.9k repo stars
▌
Service Contract Review · minimax-ai
服务/委托合同专项审查技能，由 contract-review 路由器按合同类型路由加载，不直接面向用户调用。覆盖场景：各类服务合同（咨询、运营、外包、维保、营销、设计、人力外包）、委托合同、含 SLA 服务等级约定的持续服务文本、框架服务协议+工单结构的服务文件。同义场景词：服务合同审查、服务协议审查、委托合同审查、外包合同审查、咨询合同审查、SLA 审查。执行与 nda-review 同构的全链路：matter 上下文与产物去向检查、委托方/服务方立场判定、服务立场 playbook 加载或现场补齐、Scope check（服务/委托/ 行纪/中介的类型辨析与藏条款）、七类分类检查（服务范围界定、SLA 与考核、人员投入与关键人员、验收标准与流程、知识产权归属、保密、违约与解除及管辖）、按 playbook 三色分桶、输出统一 triage memo。

12.9k repo stars
▌
Minimax Token Plan · minimax-ai
Use MiniMax Token Plan and optional pay-as-you-go multimodal tools for web search, image understanding and generation, speech, video, lyrics, music, and music covers while preserving entitlement and cost boundaries.

12.9k repo stars
▌
Skill Auto Select · minimax-ai
Design a Skill (or Plugin) that an LLM agent can reliably discover, select, and invoke based on its description, with explicit selection syntax, name-collision handling, and three-layer matching. USE WHEN: authoring a new skill for a Plugin, designing skill frontmatter, deciding between structured `UserInput::Skill` vs implicit `$skill-name` mention, handling duplicate skill names, picking between path-precise and name-based matching, or any task involving "make my skill actually get picked up by the agent". TRIGGER PHRASES: "skill selection", "skill auto-pick", "$skill-name mention", "skill description", "skill metadata", "SkillMetadata", "ExplicitSkillLookup", "three-layer matching", "name collision", "ambiguous skill name". SKIP WHEN: writing a one-shot script (use `error-recovery-strategy` or similar task skill), skill is human-only (no agent invocation), skill is bundled and not selectable.

12.9k repo stars
▌
Employee Handbook Review · minimax-ai
当用户提供规章制度、员工手册、考勤制度、奖惩制度或纪律处分办法文本，要求审查其民主程序、公示告知与内容合法性时使用。覆盖场景：新建制度合规审查、既有制度补正评估、依制度拟作违纪解除前的制度有效性核查、制度与劳动合同冲突排查。同义场景词：员工手册审查、规章制度民主程序、制度公示、违纪解除依据审查。执行链路：民主程序要件核查、公示告知证据评估、内容合法性审查、与劳动合同冲突处理、三色分桶输出补正流程建议，收尾过引用审计。

12.9k repo stars
▌
Patent Disclosure Review · minimax-ai
当用户提供技术交底书、发明提案或研发成果说明，需要评估可申请专利性、选择专利类型、检查交底书完整性时使用。覆盖场景：技术交底书提交前审查、发明点提炼、发明/实用新型/外观类型选择、新颖性创造性初筛、申请前保密风险提示。同义场景词：技术交底审查、专利交底书、发明提案评估、可申请专利性评估。执行链路：发明点提炼、类型选择、新颖性创造性初筛（检索人工执行）、交底书完整性五要件核对、申请前公开风险排查、输出审查意见，收尾过引用审计。本技能不代写权利要求、不承诺授权。

12.9k repo stars
▌
Retry With Backoff · minimax-ai
Execute explicit retry policy: max 3, base 2s, max 30s, full jitter, 60s total budget, respects `Retry-After`. USE WHEN: `error-recovery-strategy` classified error as `transient` and chose `retry`, HTTP 429 with `Retry-After` header, network timeout/refused/reset, queue/lock/eventually-consistent read returned stale, user said "重试" / "retry" / "再试" / "等一下" / "等几秒" / "backoff" / "exponential" / "rate limit" / "429" / "限流". TRIGGER PHRASES: "重试", "retry", "再试", "等一下", "等几秒", "backoff", "exponential", "rate limit", "429", "Retry-After", "throttled", "限流", "busy", "服务忙". SKIP WHEN: error is `deterministic` (won't change on retry), error is `unknown` (escalate to ask-user), work is time-sensitive and 30s backoff is too late.

12.9k repo stars
▌
Tool Output Budget · minimax-ai
Truncate oversized tool output so it does not blow the agent's context window. USE WHEN: tool output > 3000 tokens, line > 500 chars, large log, JSON array, minified code, fetched HTML, verbose npm/cargo/test output, `cat` of a big file, "truncated" / "output cut off" / "use offset to read more" message. TRIGGER PHRASES: "输出太长", "context 满了", "log 太大", "截断", "truncate", "output cut off", "读不完", "太大了", "context 撑爆", "too long". SKIP WHEN: output is small (<100 lines), output is the user-facing final answer, output is structured and needs full parse (read once with a guard).

12.9k repo stars
▌
MCP Server Patterns · minimax-ai
Reference patterns for designing, evaluating, integrating, and debugging MCP servers. Use this Skill whenever the task involves deciding whether to add an MCP server, reviewing a server design, integrating one into a workflow, or diagnosing why one fails.

12.9k repo stars
▌
Session Branch Fork · minimax-ai
Design a session-level fork / revert / recover / suspend mechanism over a paginated history with lineage tracking, immutable segments, global lock, ModelContext reconstruction, and bounded replay. USE WHEN: designing session persistence, building a "fork this conversation" feature, building "undo last N turns", building "suspend and resume later", implementing a paginated history with segment-level cursor, reconstructing ModelContext from disk, or any task involving "session as a git-like object graph". TRIGGER PHRASES: "session fork", "session branch", "thread fork", "thread rollback", "revert thread", "ThreadRollback", "SuspendTurnAndShutdown", "Op::RecoverTurn", "paginated history", "RolloutLineage", "ForkBoundary", "RolloutReferenceIndex", "ModelContext reconstruction", "ReverseJsonlScanner", "bounded replay", "git baseline", "writer lock", "subagent lineage". SKIP WHEN: single-session task state (use `world-state-tracking`), no need to undo, no need to fork.

12.9k repo stars
▌
Technology Contract Review · minimax-ai
技术合同专项审查技能，由 contract-review 路由器按合同类型路由加载，不直接面向用户调用。覆盖场景：技术开发合同（委托开发/合作开发）、技术转让合同（专利权/专利申请权/技术秘密转让）、技术许可合同（专利实施许可/技术秘密使用许可）、技术咨询与技术服务合同、含里程碑验收的研发合作文本。同义场景词：技术开发合同审查、技术转让审查、技术许可审查、研发合作协议审查、技术合同把关。执行与 nda-review 同构的全链路：matter 上下文与产物去向检查、委托方/受托方（让与方/受让方、许可方/被许可方）立场判定、技术合同立场 playbook 加载或现场补齐、Scope check（开发/转让/ 许可/咨询/服务的类型先定与藏条款）、七类分类检查（技术目标与里程碑验收、成果归属与后续改进、技术资料交付与保密、知识产权侵权担保与责任、价款与提成、违约与解除、管辖与争议解决）、按 playbook 三色分桶、输出统一 triage memo；涉外技术进出口管制因素触发升级。

12.9k repo stars
▌
Binder Replay · minimax-ai
当需要回放已记录的 binder 设计 campaign、重现每一步"当时知道什么、决定了什么"时使用；对应 openbinder replay（--fixture/--events/--source）。典型触发语："帮我回放一下 IL-7Ra 的 binder campaign""回放那个失败的 campaign""看看当时每一步都知道了什么"。同义场景：campaign replay、事件流回放、replay report、gap 分析、时间线重建。

12.9k repo stars
▌
Binder Report · minimax-ai
当需要把 output/ 下已产生的 replay、agent 评测、ranking 审计产物汇总成最终报告束时使用；收集既有 artifact、按六类证据分级标注、生成 limitations 并提交 stage-gate 审批。典型触发语："把这些产物汇总成最终报告""出一份 binder 最终报告和 limitations""把 replay/评测/审计结果打包成报告束"。同义场景：最终报告、报告束、evidence grading、limitations、claim check、收尾报告。

12.9k repo stars
▌
Binder Source · minimax-ai
当需要登记、查看、锁定或按白名单下载 binder 数据源文件时使用；对应 openbinder source list/inspect/lock/download 四个子命令。典型触发语："看看 anthropic-claude-binder-v1 这个源""inspect 一下远端文件列表""把源锁定到当前 revision""下载那个 README 小文件"。同义场景：数据源注册、source registry、pin revision、source.lock.json、下载白名单、download policy、blocked 文件、源不可用怎么办。

12.9k repo stars
▌
Goal Token Budgeting · minimax-ai
Track running token usage against goal's `token_budget`, surface at 50/80/100% thresholds, stop at 100%. USE WHEN: `goal-persistence` active AND user provided `token_budget`, user said "do X within Y tokens" / "用 Y token 完成" / "不要超预算" / "stayed within budget" / "超出预算" / "用了多少 token", about to start sub-task and need to know remaining budget, at every compact / turn boundary. TRIGGER PHRASES: "token budget", "预算", "Y tokens", "不要超过", "stayed within budget", "超出预算", "用完没", "用了多少 token", "预算跟踪", "50% / 80% / 100%", "token 预算". SKIP WHEN: goal has no budget (user did not set one), user explicitly said "no budget tracking for this one".

12.9k repo stars
▌
Plugin Author Helper · minimax-ai
Design, validate, and ship a marketplace Plugin (or Skill bundle) with proper manifest format, multi-ecosystem compatibility, version pinning, manifest fallback, install idempotency, and three-layer startup sync. USE WHEN: writing a new Plugin manifest, picking manifest format (Legacy vs AgentPlugin), adding `skills` / `mcp_servers` / `apps` / `hooks` / `interface` fields, validating a plugin before publish, designing marketplace install/remove/upgrade flows, or any task involving "make my plugin actually work in Codex". TRIGGER PHRASES: "plugin manifest", "PluginManifest", "marketplace", "agent plugin", ".agents/plugins/marketplace.json", "manifest fallback", "idempotency key", "plugin author", "plugin publish", "plugin version", "startup sync", "lock file". SKIP WHEN: writing a single skill (use `skill-auto-select`), pure MCP server (use `mcp-server` directly), one-off tool without packaging.

12.9k repo stars
▌
World State Tracking · minimax-ai
Track running state of long task in a single dedicated file that survives compaction. USE WHEN: task is long, agent has lost thread, user asks "where are we" / "到哪了" / "我们到哪了", before `context-pressure-compact`, `todowrite` alone is too thin, agent has done > 10 tool calls, "lost the thread" / "继续" / "忘了". TRIGGER PHRASES: "where are we", "到哪了", "我们到哪了", "继续", "lost thread", "忘了", "lost the thread", "我们刚才说到哪了", "走神了", "回到主线". SKIP WHEN: short task (<5 tool calls), single one-shot question, "do X" with X being small.

12.9k repo stars
▌
Binder Handoff · minimax-ai
当需要把 binder 设计算力任务打包交给外部平台（Claude Science 或自有服务器）、或校验/导入外部返回的结果包时使用；对应 openbinder handoff export/validate/import。典型触发语："把这个 target 打包成 handoff 任务包""校验一下对方返回的 result package""把外部算回来的结果导入进来"。同义场景：external handoff、任务包导出、结果包校验、回传包导入、RUNBOOK。

12.9k repo stars
▌
Delegate With Context · minimax-ai
Hand off a sub-task to a sub-agent with a tight, complete brief in the `prompt` — not the full conversation history. Apply the 4-part message envelope (Task name / Sender / Task / Payload + return path). USE WHEN: about to call `task()` to hand off a sub-task, the full conversation history is too large to forward, a minimal-context brief would do, the previous sub-agent failed because the brief was incomplete. TRIGGER PHRASES: "delegate", "hand off", "sub-agent", "delegate this", "delegate to", "派给", "委派", "让 sub-agent 干", "把 ... 交给 ...". SKIP WHEN: the sub-task is so trivial a `read` will do, you are about to do the work yourself, the user explicitly wants you (not a sub-agent) to do it.

12.9k repo stars
▌
Fork Context Decision · minimax-ai
Decide how much parent context to include in a sub-agent's `prompt` before spawning it. Pick "all / N turns / brief only" explicitly, not by accident. USE WHEN: about to call `task()` to hand off work, designing a multi-agent flow, sub-agent failed and debugging whether cause was over- or under-forking, user said "give it the full history" / "no history" / "just the brief" / "don't carry context" / "深度 fork" / "不要带 context". TRIGGER PHRASES: "fork 深度", "give it the full history", "深 fork", "no history", "just the brief", "不要带 context", "fork 0", "fork all", "完全独立会话", "轻量 context". SKIP WHEN: sub-task is trivial (one-line read), you have already decided "no context" (no decision to make).

12.9k repo stars
▌
Shareholder Agreement Review · minimax-ai
当用户需要审查股东协议、合资合同、增资协议中的股东权利安排时使用。覆盖场景：投资入股前协议审查、创始股东间协议、合资经营企业合同、股权结构调整配套协议、对赌与回购条款效力评估。同义场景词：股东协议审查、合资合同审查、SHA review、投资协议审查。执行全链路：投资方/创始方（买方/卖方）视角选择、出资与股权结构、治理席位、一致行动与表决权委托、股权转让限制（优先购买权/随售/拖售）、回购与对赌效力边界、僵局处理、违约与退出逐项检查，🟢🟡🔴 三色分桶 memo；不硬编码实体立场。

12.9k repo stars
▌
Binder Protocol · minimax-ai
当需要把 binder 设计协议草稿编译成结构化 protocol.yaml、校验协议、或按阶段切出 stage slice 时使用；对应 openbinder protocol validate/compile/show 三个子命令。典型触发语："把这份 protocol 草稿编译一下""校验这个 protocol.yaml""把 cofold_scoring 阶段的切片拿出来"。同义场景：协议编译、protocol compiler、十阶段、stage slice、阶段上下文、protocol validate。

12.9k repo stars
▌
Tool Discovery Pattern · minimax-ai
Design a tool that an LLM agent can reliably discover, search, and invoke — with proper schema, defer_loading, two-dimensional type classification, OpenAI protocol compatibility, and a tool-suggestion approval flow. USE WHEN: writing a new tool for an agent, designing the JSON schema for a tool, deciding between Function / Freeform / Namespace, fixing MCP tools that don't work with OpenAI models, building a tool-search index, designing a "request plugin install" flow, or any task involving "make my tool actually get picked up by the agent". TRIGGER PHRASES: "tool discovery", "tool search", "tool spec", "DiscoverableTool", "defer_loading", "tool_suggestion", "request_plugin_install", "MCP tool", "Dynamic tool", "JSON schema for tool", "responses API tool", "ResponsesApiFunctionTool", "ResponsesApiCustomTool", "ResponsesApiNamespace". SKIP WHEN: writing a Skill (use `skill-auto-select`), building a plugin manifest (use `plugin-author-helper`), single-use CLI script (not a tool).

12.9k repo stars
▌
Error Recovery Strategy · minimax-ai
Classify error into 4 buckets (transient / deterministic / stale / unknown) and pick one of 5 actions (retry / switch / fallback / refresh-then-retry / ask-user / skip). USE WHEN: tool returns non-success, sub-agent `status: closed-failed`, exception escapes, timeout fires, weird partial-success result, ECONNREFUSED / 5xx / 429 / timeout / permission denied / "command not found" / "fail" / "error" / "出错了" / "挂" / "失败". TRIGGER PHRASES: "出错了", "failed", "挂", "error", "失败", "fail", "permission denied", "command not found", "ECONNREFUSED", "timeout", "挂了", "再试一次", "retry", "这不行", "没用", "fallback", "退路", "不行", "跑不通", "broken". SKIP WHEN: operation succeeded, error is in user input (clarification case), error is part of expected flow (grep 0 matches).

12.9k repo stars
▌
Streaming Output Reader · minimax-ai
Read long streaming responses in bounded chunks with cumulative summary, max 3 reads, never loop. USE WHEN: tool returns long stream (SSE / WebSocket / `tail -f` / large log), output might be > 3000 tokens, file size unknown, previous read returned "truncated" / "use offset to read more" / "output cut off", `tail` of a growing log, "流式" / "实时" / "incremental" / "read in chunks". TRIGGER PHRASES: "流式", "streaming", "实时", "tail -f", "real-time", "一边跑一边看", "log 在长", "output cut off", "读到一半卡了", "incremental", "stream-read", "read in chunks", "流式读取". SKIP WHEN: output is small (<100 lines), output is structured and needs whole parse, polling for specific event (different pattern).

12.9k repo stars
▌
Binder Agent Eval · minimax-ai
当需要在相同的阶段输入上评测 LLM agent 的结构化决策、或对比多个 provider 的决策质量时使用；对应 openbinder evaluate-agent（--provider/--compare/--case）与 openbinder provider doctor。典型触发语："用 mock provider 跑一遍 decision case 评测""对比一下几个 provider 在同一批 case 上的表现""先 doctor 一下 minimax 配置通不通"。同义场景：agent 评测、decision evaluation、visible/hidden 分层、泄露检查、provider 对比、冻结决策。

12.9k repo stars
▌
Context Pressure Compact · minimax-ai
Compress a long-running multi-step task into a structured snapshot before continuing. USE WHEN: `todowrite` > 5 items, after ~20 tool calls, context getting full, agent has lost track of goal, user said "compact" / "summarize" / "refocus" / "压缩" / "总结" / "到哪了", before context window fills (>80%), before `context-pressure-compact` boundary. TRIGGER PHRASES: "compact", "summarize", "refocus", "压缩", "总结", "到哪了", "context 满了", "忘了目标", "we're getting lost", "compress", "snapshot". SKIP WHEN: short task (<5 tool calls), user in middle of dictating a request, user said "do not summarize" / "keep everything".

12.9k repo stars
▌
Subagent Family Tracking · minimax-ai
Track parent/child thread tree of spawned sub-agents with Open/Closed status. USE WHEN: spawned one or more sub-agents, task description suggests a tree (sub-tasks, "for each of A/B/C", "5 stages"), user asks "what's your sub-agent doing right now" / "子 agent 都在干嘛", sub-agent may fan out, want to know "还在跑吗" / "还有几个没关", before declaring fan-out done (all children closed check). TRIGGER PHRASES: "子 agent 都在干嘛", "sub-agent", "子任务", "family tree", "who is running", "还在跑吗", "还有几个没关", "what's your sub-agent doing", "subagent family", "subagent tree", "all children closed". SKIP WHEN: sub-task is so cheap you'd just inline it, harness already exposes live sub-agent dashboard, you are the child not the parent.

12.9k repo stars
▌
Binder Ranking Audit · minimax-ai
当需要重算并审计 binder campaign 的候选排序、核对 recorded 与 recomputed 是否一致时使用；对应 openbinder audit-ranking（--fixture/--source）。典型触发语："审计一下这个 campaign 的排序""recorded 和 recomputed 对得上吗""重算 top-N 命中率和 Spearman"。同义场景：ranking audit、排序审计、ensemble 重算、hit rate、缺失数据处理、missing data 报告。

12.9k repo stars
▌
Minimax Code Trajectory · minimax-ai
Inspect or visualize privacy-aware timelines for local MiniMax Code sessions. Use when the user asks to show, draw, open, review, or diagnose a task trajectory, summarize tool or token activity, inspect compaction or failure events, or compare recent session metadata without opening raw ledger files.

12.9k repo stars
▌
Dida365 · minimax-ai
Manage 滴答清单 (Dida365 China) tasks, lists, habits, focus records and countdowns through the official Dida365 MCP server at mcp.dida365.com. Use when the user asks to check, plan, create, update, complete or review tasks and habits on dida365.com, or asks about focus history and countdowns.

12.9k repo stars
▌
Hello Mcode · minimax-ai
Verify that an Agent Plugin Skill is available. Use when the user asks to test the hello-mcode example plugin.

12.9k repo stars
▌
Ticktick · minimax-ai
Manage TickTick (international, ticktick.com) tasks, lists, habits, focus records and countdowns through the official TickTick MCP server at mcp.ticktick.com. Use when the user asks to check, plan, create, update, complete or review tasks and habits on TickTick, or asks about focus history and countdowns.

12.9k repo stars
▌
Tool Map · minimax-ai
Cross-platform inventory of CLI tools, scripts, and MCP servers installed on the user's machine. Use when the user asks what is installed, where a tool lives, or how to run something - read the cached summary first instead of re-walking the filesystem. Refresh the catalog with the bundled scan.mjs only when the user asks, just installed a tool, or the cached summary is missing a tool the user mentions.

12.9k repo stars
▌
Microbe · minimax-ai
Use when the user has a 16S or ITS feature table plus taxonomy and metadata and needs alpha/beta diversity, composition bars, differential abundance, co-occurrence networks, random-forest biomarkers, or a taxonomy tree. Drive the microbe MCP tools with local Rscript.

12.9k repo stars
▌
Article2tasks · minimax-ai
将技术周报（批量）或单篇文章整理为滴答清单待办事项，自动分类到已有清单并打标签，全程通过 MCP 写入，跨平台无需任何本地脚本。当用户提供周报内容、文章链接，或提及「周报」「文章汇总」「添加到滴答」时，使用此技能。

12.9k repo stars
▌
Hello Hooks · minimax-ai
Verify that the experimental io.minimax.mcode Hook entry of the hello-mcode-hooks plugin is wired up. Use only when the user explicitly asks to test the hello-mcode-hooks example.

12.9k repo stars
▌
Customize · minimax-ai
当用户想修改研究画像或共享守则（guardrails）时使用：换了研究领域、换了目标期刊、换了引用格式、换了算力环境、合规要求变化、想调整某条 guardrail 的严格程度。同义触发场景：改一下画像、更新我的研究信息、修改 CLAUDE.md、调整设置、换引用格式、customize、个性化配置、改规则、重新配置工作台。本技能是画像与守则的唯一维护入口：读取 CLAUDE.md、与用户逐项确认修改点、写回文件、并明确告知哪些 skill 的行为会因此变化。

12.9k repo stars
▌
Hpc Slurm · minimax-ai
当用户需要向 Slurm 集群提交计算作业，说"提交集群任务""写个 sbatch 脚本""作业还在排队吗""批量跑一组参数""作业超时被杀了""GPU 作业怎么交"时使用。本技能规定 Slurm 集群的统一操作规程：sbatch 脚本模板（分区/时长/内存/GPU 指令 + set -euo pipefail + 环境导出 + 结束自动 rsync 产物）、squeue/sacct 状态查询口径、数组任务批量提交、超时与重排队策略、产物抓回与 record_run 登记。同义触发场景：Slurm、HPC、集群、超算、排队、批处理、job array、机时。连接与主机认知纪律沿用 remote-compute（别名、画像合同），本技能只规定作业生命周期。

12.9k repo stars
▌
Paper Read · minimax-ai bundle
当需要对单篇论文做全文精读与结构化证据提取时加载：获取并解析全文文本、按章节组织阅读、逐字摘录 quote 并定位页码、产出 EvidenceItem 与阅读笔记。通常由 literature-search / literature-survey 在「检索→阅读→综述」链路中加载，或经 review-writing 链路回读证据时使用，不直接面向用户。同义场景：全文精读、证据提取、逐字摘录、阅读笔记、PDF 文本提取、 abstract-only 标注、EvidenceItem 生成。

12.9k repo stars
▌
R Analysis · minimax-ai
当用户使用 R 做统计分析、流行病学建模、生物信息分析、ggplot 作图，说"用 R 跑一下""写个 R 脚本""rmarkdown 出报告""quarto 渲染""R 环境怎么管"时使用。本技能规定本地 R 分析的统一操作规程：renv 锁定依赖环境、脚本化非交互执行（Rscript 一条命令跑完）、sessionInfo() 必须记录进 provenance、rmarkdown/quarto 作为报告产物、产物落盘 output/<skill>/<slug>/latest/ 并与 record_run 集成。同义触发场景：R 分析、R 脚本、tidyverse、ggplot、RStudio、knitr、renv、R 统计。

12.9k repo stars
▌
Stage Gate · minimax-ai
当研究流程的任一阶段产出物完成、需要用户审批后推进时使用；通常由 research-lifecycle 在每个阶段末尾自动调用，用户也可以直接说"看看当前阶段状态""审批一下""approve""revise""reject""阶段门禁""stage gate"触发。本技能负责：把阶段产出物落盘、生成 stage-report.md（本阶段做了什么/关键结果/风险与疑虑/建议）、更新 output/<project>/stage.yaml，然后停止等待用户三选一——approve 进入下一阶段、revise 携带意见重跑本阶段（原产物归档不覆盖）、reject 终止并写结题说明。同义触发场景：阶段确认、继续下一步、打回重做、终止项目、恢复进度、上次进行到哪了。

12.9k repo stars
▌
Mcode Island · minimax-ai
Push the user's terminal out of focus to a Windows desktop Dynamic Island pill so the user can watch your work without switching back to mcode. On mcode 0.2.4+ with the `io.minimax.mcode` Hooks extension enabled (forward-compatible with MiniMax-Code-Plugins PR

12.9k repo stars
▌
Skill Bridge · minimax-ai
Convert an openclaw (or similar) skill folder into a mavis/mcode-compatible skill via the bundled stdio MCP server `skill-bridge`. Use when the user wants to migrate a skill from openclaw, reuse a skill from another framework, or port a hand-written skill that does not follow the mavis schema. Do NOT use to create a brand-new skill from scratch (use `skill-creator` instead), or to lint/refine an existing mavis skill (use `skill-refiner`).

12.9k repo stars
▌
Ip Review · minimax-ai
知识产权统一入口。触发场景：用户提出商标、专利、IP 条款或软著相关的准备与审查请求时——同义场景词包括「商标注册」「商标检索」「商标近似」「品牌注册」「技术交底」「发明交底书」「专利交底审查」「能不能报专利」「IP 归属」「职务发明」「职务作品」「知识产权转让许可」「软著登记」「软件著作权」「软著材料」。用户给出文件路径、粘贴文本或口头描述场景均可。本技能是路由器：先做执业画像检查，再按docs/scenes/ip-cn.md 的 B2 路由表识别任务类型并征得用户确认，随后加载对应专项技能（trademark-search-prep、patent-disclosure-review、ip-assignment-review、 software-copyright）执行，输出统一格式的审查/准备 memo。逐项深度工作由被路由的技能完成，本技能不直接产出审查意见，不持实体立场。

12.9k repo stars
▌
Claim Check · minimax-ai
当需要检查论文草稿中的事实性论断是否有证据支撑、做引用-论断对齐审计、找出没有出处的陈述、核对文内引用是否真能支撑所在句子时加载使用；通常由 evidence-loop 或审稿类流程调用，不直接面向用户。同义场景：论断支撑检查、claim 支撑审计、无支撑断言排查、证据覆盖检查、引用与论断对齐、"这段话有没有文献撑""哪些说法没有引用"。

12.9k repo stars
▌
Epi Writing · minimax-ai
当用户需要撰写暴发调查报告、流行病学观察性研究论文、公卫项目总结、对照 STROBE 清单自查论文、按中文期刊格式整理稿件或按 GB/T 7714 排版参考文献时使用。同义场景：暴发调查报告模板、疫情调查报告、现场流行病学报告、观察性研究写作、STROBE 核对清单、队列/病例对照/横断面研究论文、中华流行病学杂志投稿、中文期刊格式、GB/T 7714 参考文献、伦理声明写作、"帮我把这次暴发调查写成报告"。

12.9k repo stars
▌
Run Monitor · minimax-ai bundle
当用户有耗时较长的本地任务需要后台执行，说"后台跑一下""这个任务要跑很久""帮我盯着这个任务""任务跑完了吗""看看日志"时使用。本技能定义长任务 Run 抽象（核心设计：不让人机等任务、轮询不耗 token）：run 目录 runs/<run_id>/{run.json, stdout.log, stderr.log}，run.json 记录 {id, cmd, cwd, pid, started, status, exit_code}；配套纯标准库脚本 run_task.py 提供 start（detached 后台启动，Windows/POSIX 跨平台）、status（pid 存活检查 + 退出码）、list、tail 四个子命令。同义触发场景：长任务、后台运行、任务监控、看进度、训练还没跑完、批处理脚本。

12.9k repo stars
▌
Spatial Epi · minimax-ai
当用户需要比较分地区发病率/死亡率、计算标准化发病比 SMR、处理小区域率不稳定问题、检测疾病空间聚集性、制作分级统计地图（choropleth）、评审一张疾病地图的规范性时使用。同义场景：空间流行病学、疾病地图、 SMR 标化、间接标化、小区域估计、率平滑、Moran's I 莫兰指数、空间自相关、扫描统计量、SaTScan、GeoDa、聚集性分析、热点分析、choropleth 制图规范、 "帮我算算各县 SMR""这病有没有空间聚集"。

12.9k repo stars
▌
Nda Review · minimax-ai
保密协议（NDA）专项审查技能，由 contract-review 路由器按合同类型路由加载，不直接面向用户调用。覆盖场景：单方保密协议、双方互负保密协议、Confidentiality Agreement、Non-Disclosure Agreement、保密条款占主体的合作前期文件。同义场景词：NDA 审查、保密协议审查、保密条款审查、confidentiality review。执行全链路：matter 上下文检查、产物去向检查、披露方/接收方/双方立场判定、 NDA 立场 playbook 加载或现场补齐、Scope check（识别名为 NDA 实为竞业限制/ 排他合作/技术许可/不招揽的藏条款）、六类分类检查清单、按 playbook 三色分桶、输出 triage memo 与 next-steps 决策树，收尾登记续期并过引用审计。

12.9k repo stars
▌
Paper Search · minimax-ai bundle
当需要真正执行文献数据库 API 检索时加载：构建检索式（布尔组合、字段限定）、调用 OpenAlex / Crossref / arXiv 接口、遵守请求礼仪（限速、UA、指数退避）、把返回解析为统一 PaperDocument。通常由 literature-search 调用，不直接面向用户。同义场景：检索执行、API 查询、文献接口调用、query 构建、检索脚本运行、接口限速与重试、检索结果解析。

12.9k repo stars
▌
Case Verify · minimax-ai
当用户需要核验案例引用——"这个案号是真的吗/查一下这个判例/指导性案例第几号还有效吗/这个案例引用靠不靠谱"时使用；也被其他技能在填入案例类 [CITE:__] 占位前调用。先校验案号格式，再按凭证配置经北大法宝/聚法查证案例存在性与基本信息；未配置商业源时诚实标注 [模型知识—待核实] 并给出人工核查路径，绝不凭记忆确认案号。同时提醒引用性质检查：引的应是裁判要旨而非顺带意见。

12.9k repo stars
▌
Cn Literature · minimax-ai bundle
当用户需要检索或整理中文文献（CNKI 知网、万方、维普、超星）、把中文数据库导出的题录文件解析成统一 PaperDocument、或把中文文献与英文检索结果合并去重时使用。同义场景：中文文献检索、知网导出题录解析、万方检索、Refworks/EndNote 格式转换、"帮我把这份 CNKI 导出的 txt 整理成文献表""中文核心期刊上关于 X 有哪些研究"。

12.9k repo stars
▌
Evidence Loop · minimax-ai
当用户希望一次性修好论文的证据链——先核验引用、再查正文支撑、对问题条目定向补检文献、迭代修订直到收敛——时使用。同义场景：证据闭环、引用修复流水线、核验-补检-修订迭代、边查边补、系统性消除无支撑论断、 "帮我把这篇文章的引用问题全部查出来并补上文献""修到没有硬伤为止"。

12.9k repo stars
▌
Seir Modeling · minimax-ai bundle
当用户需要用仓室模型做传染病传播情景推演、选择 SIR/SEIR/SEIRS 模型结构、理解 β/σ/γ/R0 参数含义、做参数敏感性分析、估计峰值时间与峰值感染数、解读或评审一篇仓室模型结果时使用。同义场景：SEIR 模型、SIR 模型、仓室模型、传播动力学建模、基本再生数 R0、有效再生数、传染期/潜伏期参数化、疫情趋势情景模拟、干预效果推演、"帮我跑一个 SEIR 看看峰值"。

12.9k repo stars
▌
Search First · minimax-ai
Enforce a search-before-edit workflow for code changes. Use this Skill whenever a task involves changing code, tracing behavior, or understanding an unfamiliar area, so the agent searches and reads the relevant files before proposing or implementing edits.

12.9k repo stars
▌
Rnaseq Plot · minimax-ai
Use when the user already has an RNA-seq count or expression matrix and needs downstream plots or analysis (normalize, PCA, DESeq2/edgeR/limma, volcano, heatmap, GO/KEGG, GSEA, WGCNA). Call the rnaseq-plot MCP tools named rgraph_*; do not redraw in Python.

12.9k repo stars
▌
Web Research · minimax-ai bundle
Use when the user asks to research a topic on the web, gather information from multiple sources, do a literature scan, summarize recent developments on a subject, or look up articles. Triggers on requests like "调研 X", "找几篇关于 Y 的文章", "research Z", "汇总 Q 的资料", "look up the latest on W". Do NOT use for single-URL fetches with no research framing (use the defuddle skill directly), local file operations, code edits, or tasks that do not need web data.

12.9k repo stars
▌
Comfyui Drama · minimax-ai
Generate a complete short drama video from a storyboard. Use this Skill for scenarios 5–6 from `comfyui-studio`'s trigger index: 5=首帧 (drama first frame, two-character LoRA + prompt → 16:9 still) and 6=出片 (image-to-video, first frame + motion prompt → MP4 clip). The Skill also describes the full 7-stage pipeline around them (storyboard → TTS → script refinement → first frame → image-to-video → subtitle burn → audio/video assembly). Ships two preset workflow templates: `workflows/drama-first-frame.json` and `workflows/drama-image-to-video.json`.

12.9k repo stars
▌
Labor Review · minimax-ai
劳动用工统一入口。触发场景：用户提出劳动合同、解除裁员、劳动仲裁或规章制度相关的审查与评估请求时——同义场景词包括「审劳动合同」「劳动合同审查」「Offer 审查」「续签变更」「辞退」「裁员」「解除评估」「经济补偿」「离职谈判」「劳动仲裁」「仲裁应诉」「仲裁请求」「员工手册」「规章制度审查」「考勤奖惩制度」「民主程序」。用户给出文件路径、粘贴文本或口头描述场景均可。本技能是路由器：先做执业画像检查，再按 docs/scenes/labor-cn.md 的 B2 路由表识别任务类型并征得用户确认，随后加载对应专项技能（labor-contract-review、termination-assessment、 labor-arbitration-prep、employee-handbook-review）执行，输出统一格式的审查/评估 memo。逐条深度审查由被路由的技能完成，本技能不直接产出审查意见，不持实体立场。

12.9k repo stars
▌
Lease Review · minimax-ai
租赁合同专项审查技能，由 contract-review 路由器按合同类型路由加载，不直接面向用户调用。覆盖场景：房屋租赁合同（办公、商铺、厂房、仓储、住宅）、设备与车辆租赁合同、场地使用协议、含装修免租期安排的租赁文本、转租与分租协议。同义场景词：租赁合同审查、房屋租赁审查、租房合同审查、设备租赁审查、租约把关。执行与 nda-review 同构的全链路： matter 上下文与产物去向检查、出租方/承租方立场判定、租赁立场 playbook 加载或现场补齐、Scope check（名为租赁实为买卖/融资/承包经营的藏条款）、八类分类检查（租赁物与用途、期限、租金与押金、维修义务、转租限制、买卖不破租赁与优先购买权、装修添附、解除与争议解决）、按 playbook 三色分桶、输出统一 triage memo。

12.9k repo stars
▌
Remote Compute · minimax-ai
当用户需要在 SSH 远程服务器上跑计算任务，说"连服务器跑一下""远程跑个任务""把代码传到服务器""服务器上有 GPU 吗""rsync 抓回结果""SSH 怎么配"时使用。本技能规定远程算力的统一操作规程：连接一律走 ~/.ssh/config 别名加 ControlMaster 共享连接复用（避免集群双因子认证反复打扰）、首次接入新主机做有界只读探测并把能力画像写入 .openscience/hosts/<alias>.yaml 当作合同（画像为 null 的能力禁止臆测）、明确的失败分支处理、远程产物必须 rsync 抓回并 record_run（不记录的运行等于不存在）、安全红线（不在命令行传密钥、不在远端存凭证）。同义触发场景：远程计算、SSH、服务器、算力、集群登录、scp、同步结果。

12.9k repo stars
▌
Review Writing · minimax-ai
当用户需要基于文献证据撰写综述、写论文引言或相关工作、把文献调研结果组织成规范成文、生成带 GB/T 7714 参考文献的综述初稿时使用。同义场景：综述写作、文献综述成文、相关工作撰写、调研报告写作、生成参考文献列表、 "帮我写一篇关于 X 的综述""把分析结果写成文章""整理成 GB/T 7714 格式"。

12.9k repo stars
▌
Comfyui Studio · minimax-ai
Drive a local ComfyUI 8188 server for image and short-clip generation. Use this Skill when the task involves a ComfyUI workflow, a local Stable Diffusion / Flux / Z-Image checkpoint, character-consistent generation with a LoRA, or producing deterministic outputs from a saved workflow JSON. Triggers on requests like "use ComfyUI to ...", "submit this workflow", "what models do I have", "generate an image with my LoRA", or "check the queue".

12.9k repo stars
▌
Due Diligence · minimax-ai bundle
当用户需要对目标公司做法律尽职调查时使用——同义场景词包括「尽调」「法律尽调」「尽职调查」「投资前帮我查查这家公司」「收购前尽调」「尽调问题清单」「帮我看看这家公司有什么法律风险」「出一份尽调报告」。本技能是法律尽调全流程编排器：matter-workspace 建档 → 按尽调目的裁剪问题清单（references/ diligence-checklist.md，确认门）→ 资料清单请求并标注缺失 → 调 tabular-review 执行 N 文档 × M 问题网格 → Flags 双轴分级汇总（红线三段式 blocks/work-but-ships/FYI）→ 尽调报告骨架成稿 → citation-audit 清零后才可发出。不下投资结论，不做财务/税务尽调，不替代律师现场核查。

12.9k repo stars
▌
Evidence List · minimax-ai
当用户需要把纠纷材料整理成可提交的证据清单时使用——同义场景词包括「整理证据」「证据清单」「证据目录」「这些材料能证明什么」「举证」「证据还差什么」「帮我列证据」。按民事诉讼证据种类（书证/物证/视听资料/电子数据/证人证言/鉴定意见/勘验笔录/当事人陈述，种类依据保持 [CITE:__] 占位）归类，逐项登记编号、名称、来源、证明目的、形式（原件/复印件/电子数据哈希）与页码，做真实性/合法性/关联性三性自查，提示举证期限与逾期后果（[模型知识—待核实]）及电子数据固定渠道，输出 evidence-list.md 表格与缺口清单（有诉请无证据的项标 🔴）。非律师使用者向法院提交前必须经执业律师复核（G5 UPL 门控）。

12.9k repo stars
▌
Matter Intake · minimax-ai
诉讼事项立案（事项接收与信息采集）。触发场景：用户考虑起诉、被起诉、收到律师函/催款函需要应对、想把一桩纠纷登记为正式事项跟进——同义场景词包括「我要起诉」「想起诉某人」「被起诉了怎么办」「打官司」「立案」「纠纷登记」「建个诉讼档案」「催款要不回来」「准备诉讼材料」。通过结构化引导采集当事人及主体资格、案由、关键事实时间线、诉讼时效初筛、管辖初筛、证据清单、对方财产线索，写入 matters/_log.yaml 登记簿与 matters/<slug>/intake.md 立案文件；时效或管辖存在明显问题时立即标记 [需复核] 并建议律师介入。

12.9k repo stars
▌
Citation Verify · minimax-ai
当用户需要核验参考文献真实性、检查 BibTeX 条目是否有误、审查论文引用是否可靠、排查 bibliography 中的错误条目、验证 DOI 与题录是否对得上、发现参考文献张冠李戴、做投稿前引用体检或审稿引用抽查时使用。同义场景：参考文献核验、引用核查、文献真伪鉴别、bib 文件检查、参考文献纠错、引用元数据比对、 "帮我看看这些参考文献有没有问题""这个 DOI 对得上吗""有没有编造的引用"。

12.9k repo stars
▌
Epi Data Access · minimax-ai
当用户需要获取流行病学与公共卫生数据、查找疾病负担或疫情公开数据源、下载国际或中国的公卫统计与监测数据、评估某个数据源能不能用/怎么引用时使用。同义场景：WHO 数据、GHDx/IHME 疾病负担、Our World in Data、ProMED 疫情快报、国家卫健委疫情通报、中国 CDC 周报、China CDC Weekly、国家统计局人口数据、 CHNS 队列、发病率/死亡率数据去哪找、疫情数据下载、公卫数据源推荐、个案数据去标识化、涉及人的数据要不要伦理审查、"帮我找找某地某病的数据"。

12.9k repo stars
▌
Python Analysis · minimax-ai bundle
当用户使用 Python 做科研数据分析、统计建模、机器学习实验或批量数据处理，说"用 Python 跑一下分析""做个统计""清洗数据""画结果图""跑个脚本"时使用。本技能规定本地 Python 分析的统一操作规程：环境优先 uv/venv 隔离、依赖写入 requirements.txt 并通过 record_run 记录环境指纹、随机种子固定并登记、原始数据只读（清洗另存新文件）、产物落盘 output/<skill>/<slug>/latest/、图表 PNG+PDF 双格式 300dpi，并附 stats_integrity_check.py 对报告数字做确定性体检。同义触发场景：Python 分析、跑数据、统计分析、数据处理、画图、pandas、matplotlib、分析环境怎么配。

12.9k repo stars
▌
Plant Public · minimax-ai
Use when the user needs plant or Solanaceae gene/protein lookup via public APIs (UniProt, NCBI, InterPro, PDB, AlphaFold, Ensembl Plants, Sol Genomics, Arabidopsis BAR/ATTED/STRING), domain annotation, or structure summaries. Do not invent sequences; call the plant-public MCP tools.

12.9k repo stars
▌
Acp Collab · minimax-ai
Use the OpenClaw-mcode-ACP inbox to collaborate peer-to-peer with goudan (OpenClaw main session) instead of one-shot task calls. Read incoming messages, push progress, ask blocking questions, and answer peer questions. Use when working on multi-step tasks that span MiniMax Code and OpenClaw sessions, when uncertain decisions need goudan's input, or when long-running work should report progress back to the parent session.

12.9k repo stars
▌
Citation Audit · minimax-ai
当任何产物（备忘录、清单、报告、合同修改建议、brief）即将对外发出——发送给相对方、提交法院/仲裁/监管机关、或用于签署——时触发执行引用审计门禁。本技能是库技能，通常由其他技能在交付前调用而非使用者直接发起：扫描 [CITE:__] 残留、核全部法条与案例引用已过 statute-verify/case-verify、检查来源标注符合词汇表、确认法域声明与保密标头档位匹配，输出 PASS/FAIL 与问题清单，FAIL 时给出逐项修复路径。

12.9k repo stars
▌
Legal Research · minimax-ai
当用户提出法律检索需求——"查一下某法规/某条还有效吗/有没有相关案例/最高法的司法解释原文/这个监管规定哪儿来的/帮我做个法律调研/美国法上怎么规定的/某问题法律上怎么说"时使用。本技能是检索路由器：先把需求分类（法律法规/司法解释/ 案例/监管规则/美国法对照），再按凭证配置选择数据源执行，所有结果强制带来源标注，未覆盖的源显式声明缺口。不直接给出法律意见，检索结论服务于后续分析。

12.9k repo stars

171 published skills · page 1 of 2