# Qeasy Cloud(轻易云) > Qeasy Cloud is an enterprise iPaaS data integration platform with 500+ prebuilt connectors and 2,000+ public documented API assets. Its flagship e-commerce financial reconciliation system supports 15 domestic and cross-border platforms with three-way matching, 23 variance reasons and posting to Kingdee, Yonyou or SAP. The Insights hub (/insights) is the core content section. Qeasy also exposes standardized, auditable capabilities to external AI agents via the Qeasy Knowledge MCP Server (/mcp) and SKILL.md-style Agent Skills (/skill). ## AI Agent Quick Start (read first) - [Read this site map instead of crawling every page](https://www.qeasy.cloud/en/llms.txt): AI agents are welcome to use this dedicated entry point. This file maps products, documentation, agent registration and MCP access. Chinese version: /llms.txt. - [Step 1: Request an agent registration challenge](https://www.qeasy.cloud/api/v1/agent-registrations/challenge): POST here to receive challengeId, algorithm, difficulty and expiry. No phone or browser session required. - [Step 2: Register the agent and receive a Standard key](https://www.qeasy.cloud/api/v1/agent-registrations): After solving SHA-256 proof-of-work, POST name/framework/version/challengeId/nonce here. The plaintext key is returned only once; store it securely. - [Step 3: Follow the dedicated agent integration guide](https://www.qeasy.cloud/en/mcp#getting-started): Includes full registration examples, MCP initialize / tools/list / knowledge retrieval flow, mcp.json and error handling. ## Platform - [Qeasy Cloud](https://www.qeasy.cloud/en): iPaaS data integration platform with 500+ prebuilt connectors for no-code ERP / e-commerce / CRM / WMS / MES integration. Flagship product: e-commerce financial reconciliation. - [Features](https://www.qeasy.cloud/en/features): Visual configuration, connector marketplace, monitoring & alerting, and reconciliation capabilities. - [Integrated Platforms](https://www.qeasy.cloud/en/platform): Directory of 500+ pre-integrated business systems — ERP, e-commerce, CRM, WMS, MES — ready to use out of the box. - [Connectors](https://www.qeasy.cloud/en/connectors): Prebuilt interface adapters for ERP, e-commerce, CRM and WMS systems; no-code configuration. - [API Documentation](https://www.qeasy.cloud/api): API market with request/response references and integration guides. - [Monitoring Center](https://www.qeasy.cloud/en/monitor): Pipeline runtime status, anomaly alerting and log tracing. ## Latest Insights Articles - [大模型时代的财务系统:RAG + 沙箱的工程哲学](https://www.qeasy.cloud/en/insights/reconciliation/7-2-4-rag-sandbox-engineering-philosophy-llm-era-financial-systems): 2026 年,CFO 桌上最常出现的一份报告,不是月报也不是预算,而是 Gartner 那份题为《2026 大模型在企业财务场景的落地现状》的白皮书。 - [业务人员如何用 AI 财务智能体快速响应平台规则变化?](https://www.qeasy.cloud/en/insights/reconciliation/5-1-5-business-user-ai-financial-agent-quick-response-platform-rule-changes): 当京东新增「营销券」、亚马逊新增「代扣税」、抖音上线「直播券」时,业务人员能不能不排开发、不写一行代码,让新规则当天生效?答案是可以。本文给业务人员一份「4 步快速响应」的实操指南:对话描述新规则 → AI 自动勘察/编写/试跑/保存 → 业务人员一句话确认 → 上线,全流程 30-90 分钟,零开发依赖。 - [小红书电商 对账预览 解析要点](https://www.qeasy.cloud/en/insights/reconciliation/4-7-2-xiaohongshu-reconciliation-preview-parse-key-points): 小红书电商账单是 5 大已落地平台(京东 POP / 抖店 / 支付宝 / 亚马逊 / 速卖通)之外的「第六张面孔」——账户在千帆商家后台导出、双账单体系(资金明细 7 列货款视角 + 结算明细 24 列订单视角)、交易类型只有「结算入账 / 退款」2 类、计佣基数口径(实付/实退 + 平台/政府券补贴)与金蝶应收一致、单个动账原始行按 6 个费用列自动拆出多条子行("一拆多")。本文把小红书双账单的字段结构、解析要点的 2 个版本演进(v1.0 → v1.1.0)、与已落地 5 平台的 3 大差异点讲透,让做小红书的财务在接入之前就了解账单长什么样、接入路径要怎么走。 - [轻易云 ERP 实施顾问 5 大操作](https://www.qeasy.cloud/en/insights/reconciliation/9-5-4-qingyi-yun-erp-implementer-5-common-operations): 作为 ERP 实施顾问,你每天面对的是金蝶云星空 / 用友 / SAP 这类外部 ERP——对账结果能不能进 ERP、转换规则对不对、推送报错怎么查、状态卡在哪一态、凌晨告警怎么定位——这 5 件事构成了你的「日常 5 大常用操作」。本文按「配置 → 测试 → 调试 → 监控 → 排查」的顺序,把每一步的入口、契约、状态机、踩坑与 B 级干货拆给你看。无论你面对的是金蝶云星空的 9 个 handler、用友 NC 的 PULL 链,还是 SAP S/4 HANA 的 OData 推送,这 5 步走完就能完成一次完整的「集成中心 + 集成转换 + 推送落地」实施闭环。 - [财务人员的"AI 提问力":3 大自然语言查询场景](https://www.qeasy.cloud/en/insights/reconciliation/5-3-4-financial-staff-ai-questioning-power-3-natural-language-query-scenarios): 同一家公司、同一个 AI 财务助手、同一组对账数据——3 个财务问出 3 种结果:A 财务问"京东订单为啥对不上"得到 18 分钟翻文档的笼统答复;B 财务问"京东 POP 9 月 22 单 AMOUNT_DIFF 按 diffReason 分布、TOP 3 业务场景、给处理建议"得到 3 秒定位 + 业务语言结论;C 财务直接问"给我个报表"得到 0.8 秒的"信息不足、无法作答"提示。**问题不在 AI,问题在「提问力」**。本文给财务从业者一份「AI 提问力」实战手册——**3 大自然语言查询场景(异常查询 / 进度查询 / 报表查询)的提问模板 × 4 件套(场景 + 维度 + 口径 + 期望) × 5 个反直觉陷阱**——让财务人员从「用工具」升级到「会提问」。**关键不是「装 1 个 AI」,而是「每个财务会用 AI 的方式问对问题」——这是 2026 年电商财务人才转型的分水岭**。 - [主数据管理(MDM)实战:5 大域 × 电商对账的统一主数据支撑](https://www.qeasy.cloud/en/insights/reconciliation/13-1-4-master-data-management-mdm-practice): 上一篇文章(数据标准与元数据管理)解决的是「数据应该长什么样、被机器怎么读到」,这一篇要把标准落到具体对象上——主数据。主数据是「跨业务、跨系统、跨部门反复复用的核心实体」,一家年 GMV 10 亿的电商企业,主数据出问题影响的不是「某一个 Excel 对不上」,而是「平台、供应链、财务、ERP 四套系统的同一件商品、同一笔客户、同一张凭证」全部错位。 - [D-V7 fire-and-forget API 启动](https://www.qeasy.cloud/en/insights/reconciliation/2-4-2-dv7-fire-and-forget-api-startup-under-1-second): API 启动 < 1 秒不是优化出来的,而是设计出来的。把知识库 embedding 回填从「启动期 await 阻塞」改为「`setImmediate` 后台 fire-and-forget」是这套架构里最反直觉却最关键的决策——`app.listen()` 必须先返回、后台补算绝不平 await、embedding 失败写 1536 维全零向量 + `status='failed'`、召回 SQL 硬过滤 `status='ready'` 即可排除失败行不误命中。这篇文章把 12 行的 `main.ts` 启动钩子和 100 行的 `EmbedderInitializer` 循环拆开,讲清楚 D-V7 启动时序决策的 6 条工程纪律、防死循环的边界设计、以及为什么「先 listen + 后台算」是 AI 工程从「能用」升级到「可预测」的关键转折。 - [财务人员的业务理解能力培养:5 步法 + 4 大场景,让财务真正"懂业务"](https://www.qeasy.cloud/en/insights/reconciliation/12-1-3-finance-staff-business-understanding-5-steps-4-scenarios): 中国注册会计师协会 2026 年 5 月调研显示,**78% 中型企业财务负责人承认"被业务部门评价为'不懂业务'"**——这是财务数字化推进受阻的最常见原因,比"工具不好用"高 2.7 倍。但只有 11% 的企业为此做过系统的能力培养。本文给出一份**5 步法 + 4 大场景**的实操路径:**5 步法**让财务从"办公桌"走到"业务现场"——参与业务会议 / 跟单 / 跟仓储 / 跟营销 / 跟客服;**4 大场景**让财务和业务部门建立共同语言——选品 / 促销 / 物流 / 售后。每一步都给具体动作、产出物、典型时间投入——给想从"记账员"走到"业务伙伴"的财务人。 - [eBay 对账:拍卖 + 一口价 + 海外仓 EIS](https://www.qeasy.cloud/en/insights/reconciliation/11-4-4-ebay-auction-buyitnow-overseas-warehouse): eBay 美区对账不是「1 个美亚分站 + 1 张卖家中心对账单」的简单结构,而是 **「拍卖(Bidding / 中标结算)+ 一口价(Buy It Now / 立买立卖)+ 海外仓 EIS(eBay International Shipping / 5 段费率)」** 三套订单范式叠加 **「拍卖佣金特惠 8% vs 一口价佣金 12% + 周结算 vs T+14~T+30 海外仓延迟放款 + 30 国币种 + 美国 50 州销售税 + Best Offer 议价 + eBay Money Back Guarantee 强制退款」** 的乘法复杂度。一笔订单的费率可能在「8% / 12% / EIS 5 段费率」3 套规则中跨越 3 个口径,账期从 T+2 跨到 T+30。本文用 5 张图拆解 eBay 对账的 3 道断层(**订单类型断层 + 资金链路断层 + 费率断层**),并给出拍卖佣金特惠判定 5 步法、EIS 海外仓 5 段费率拆解、5 表勾稽链的实操工具。 - [SAP S/4HANA 集成:新一代 ERP 的对账接入](https://www.qeasy.cloud/en/insights/reconciliation/6-4-1-sap-s4hana-integration-next-gen-erp-reconciliation): SAP S/4HANA 是 SAP 在 ECC 之后主推的新一代 ERP,2015 年首发、2023 年把 ECC 推入「延伸维护」、2027 年 ECC 标准维护彻底停止——任何还有 ECC 的大中型企业,**未来 18 个月内必须做一次 SAP S/4HANA 的选型与对接**。但 SAP S/4HANA 不是「升级」,是「重做」:① 数据模型从行存储改为列式 CDS View + RAP 编程模型,② API 体系从 BAPI/IDoc/RFC 升级为 OData V2/V4 + SAP Business Technology Platform(BTP),③ 集成从 SAP PI/PO 双堆栈迁到 SAP Integration Suite。本文用 SAP S/4HANA 与 SAP ECC 的 6 维对比表,给 CTO / ERP 实施一份完整的对账接入路径,结尾给出 4 步对接 + 5 步验收清单。 ## Products - [Qeasy E-commerce Smart Reconciliation Platform](https://www.qeasy.cloud/en/reconciliation): Production-grade e-commerce financial reconciliation system — the platform's flagship product. Connects to 15 major platforms and 500+ ERP/finance systems with three-way matching, 23 built-in variance reasons, five-dimension accounting, accrual-basis revenue recognition, one-click ERP posting, multi-currency, EU VAT and export rebate handling, and a human-confirmed AI finance assistant. - [AI Smart Replenishment](https://www.qeasy.cloud/en/replenishment): AI agent-powered replenishment decisions: three registered agents compute weekly plans over live sales, inventory and in-transit data — every suggestion backed by data, formulas and reports, fully traceable. - [Enterprise AI Integration](https://www.qeasy.cloud/en/ai): AI chat, image generation and knowledge-base LLM training for enterprises. - [Qeasy Knowledge MCP Server](https://www.qeasy.cloud/en/mcp): Retrieval MCP server for Qeasy knowledge across two domains: e-commerce reconciliation and the data integration platform (public layers searchable with Standard keys; live counts via knowledge_domains). Four tools: knowledge_search / knowledge_fetch / knowledge_domains / knowledge_reference (running-strategy reference, restricted REF layer); stateless JSON-RPC over POST /api/mcp. - [Qeasy Agent Skills (SKILL.md packs)](https://www.qeasy.cloud/en/skill): SKILL.md-style skill packs for data integration, e-commerce reconciliation and smart replenishment, distilled from production workflows. Reconciliation skills encode statement parsing, order-to-ERP matching, discrepancy triage and accrual-basis revenue recognition, with bilingual frontmatter, deterministic API-backed steps and propose-then-confirm gates. ## MCP Server - [Agent Registration (required first)](https://www.qeasy.cloud/api/v1/agent-registrations): Every MCP call requires a key. POST /api/v1/agent-registrations/challenge, solve sha256(challengeId + ':' + nonce) for the requested leading zero bits (default 18), then POST /api/v1/agent-registrations with name/framework/version/challengeId/nonce. The response issues a Standard key (20 public calls/day per account, shared by all Standard keys) and mcp.json. No phone, session or CRM lead. Plaintext is returned only once. - [MCP Endpoint & Health](https://www.qeasy.cloud/api/mcp): POST /api/mcp — strict JSON-RPC 2.0 over stateless MCP Streamable HTTP; Authorization: Bearer qk_... is required. GET /api/v1/mcp/health performs real PG/Redis/embedding probes; /api/v1/mcp/metrics requires a separate token. Responses include X-Request-ID, RateLimit, cache and timing headers. - [MCP Quickstart & Protocol](https://www.qeasy.cloud/api/mcp): Flow: initialize (protocolVersion/capabilities/clientInfo required) → tools/list (inputSchema + outputSchema) → tools/call. Body limit 64 KiB; 401 includes WWW-Authenticate; notifications return 202 with no body. Retrieval only, no generated answers. - [Tool: knowledge_search](https://www.qeasy.cloud/en/mcp): Hybrid vector + keyword retrieval with RRF ranking. Arguments: query, product/type/domain/platform, limit (1–20), cursor, minScore, granularity=doc|atom and includeMeta; data-integration knowledge adds scenario (S1/S2/S3, default S2, hard-filtered before recall) and category (K1/K2/K3). score is normalized 0–1 with vector/keyword/rrf breakdown; off-topic queries return hits:[]. Use page.nextCursor for pagination; cache keys bind authorization, filters and corpus revision. - [Tool: knowledge_fetch](https://www.qeasy.cloud/en/mcp): Fetch a document or a single chunk by docId, citation or atomId. Citation anchors differ by product domain: reconciliation = sourcePath#KA-atomId (KA-…-NNN); data-integration = sourcePath#section (Chinese heading) or sourcePath#chunkId (chk_…). maxChars caps size; doc-level responses carry chars and a sizeHint above 60,000 chars (prefer maxChars segmentation). format=json (TPL template documents only): returns structured template={source,target,scripts[]} with granularity=template (parse issues go to parseWarnings; no scripts → scripts:[] + scriptsNote); the default markdown is the cleaned template document (no duplicated headings, source/target meta_data JSON blocks included). Invalid anchors return -32602. Restricted documents and documents outside key product/type scope return -32403; known docIds cannot bypass authorization. - [Tool: knowledge_domains](https://www.qeasy.cloud/en/mcp): Returns visible product × type counts, live stats and domain/platform facets without daily quota. Public: 1421 docs / 82 atoms; authorized: 2811 docs / 811 atoms. Platform topics may describe boundaries/prerequisites, not live reconciliation support. - [Tool: knowledge_reference](https://www.qeasy.cloud/en/mcp): Running-strategy reference search (REF layer, group granularity, all restricted): filter by sourcePlatform/targetPlatform (path slug, upstream code or Chinese vendor name, case-insensitive), interface (live meta_data api, prefix or exact), object (FormId; pairs without a FormId use an "api:…" path — identified by objectType=formId|api) or field (config field name) in any combination, with an optional query for hybrid recall; sortBy=evidence|finished|recent. Each ref carries citation, quality signals (finished/errorRate = group aggregates, score = the sampled strategy's raw score, 0 = platform did not score, active = lastSyncDays ≤ 7, lastSyncDays) and a templateCitation — fetch it via knowledge_fetch (format=json returns structured {source,target,scripts}; the default markdown is the template document with the S/V/E/N checklist and source/target meta_data JSON blocks) and adapt it in the platform MCP. refs = cards on the current page (limit default 8 / max 20); candidateCount = matching group count; cursor paginates across pages. Standard keys get an empty refs list plus accessHint (restricted); an authorized empty refs list means the dimensions are not covered — not a permission failure; charged like knowledge_search with complete cache HITs free. - [MCP Citation & Response Contract](https://www.qeasy.cloud/en/mcp): Every hit carries a citation — reconciliation: docs/knowledge/.md#KA-...; data-integration: docs/knowledge/#section — and answers must cite it. Excerpts are capped at 500 characters. Response shape: content[0].text is a Markdown summary for humans/models; machine-readable data lives in result.structuredContent (hits/page/quota). All tools declare outputSchema. Evidence only, no generated prose. - [MCP Errors, Quotas & Cache Charging](https://www.qeasy.cloud/en/mcp): Missing keys return HTTP 401 with registration hints. Daily/minute limits return HTTP 429 + -32029 with errorKind=daily_quota or minute_rate_limit. Standard=20/day and 120/min per account, shared by all Standard keys; authorized=600/min. Complete cache HITs consume neither daily nor minute allowance; MISS/COALESCED/BYPASS are charged. atomId requires a -NNN suffix. -32602 = invalid params; -32601 = unknown JSON-RPC method or unknown tool name; -32403 = restricted or out of scope. - [MCP Access Tiers](https://www.qeasy.cloud/en/mcp): Humans register at /register and agents at /api/v1/agent-registrations; both automatically receive a Standard key (20 public calls/day per account, shared by all Standard keys; public only: 1421 docs / 82 atoms). Anonymous MCP is removed. Authorized access (2811 docs / 811 atoms) still requires Qeasy approval. - [Knowledge domain: data-integration / product (1416 docs, public)](https://www.qeasy.cloud/en/mcp): Public data-integration knowledge (product manuals and guides), retrievable via knowledge_search / knowledge_fetch. - [Knowledge domain: reconciliation / product (5 docs, public)](https://www.qeasy.cloud/en/mcp): Public reconciliation knowledge (product manuals and guides), retrievable via knowledge_search / knowledge_fetch. - [MCP Workflows (Blueprint + Historical Reference)](https://www.qeasy.cloud/en/mcp): Solution blueprint workflow (verified practice): narrow first with product + scenario + category + platform (platform accepts path slugs, upstream codes or Chinese vendor names, e.g. kingdee-cloud / Kingdee.Cloud / 金蝶云星空), then fetch the key sections (field mapping / incrementality / risks) instead of full documents. The corpus marks what it cannot derive (pair documents §7: “no dominant value / at most N% — must choose per customer account”) — do not fabricate. Historical reference workflow (M28): knowledge_reference finds the group card for a running strategy by platform pair / object / field → knowledge_fetch(templateCitation) pulls the copyable template (format=json returns structured {source,target,scripts}; the default markdown is the S/V/E/N checklist plus source/target meta_data JSON blocks) → paste and adapt it in the platform MCP (strategy.create/copy + source/target update) — copying a proven running strategy beats configuring from scratch. For object/pair-level knowledge, query with identifiers (interface / object / field names) — far better hit rates than business prose. For Q&A, fetch one citation#section from the top hits. - [MCP Getting Started Guide](https://www.qeasy.cloud/en/mcp): Endpoint, auth tiers, mcp.json for Claude Desktop / Cursor, raw JSON-RPC examples, tool parameters, response/citation contract, error codes and roadmap — everything needed to connect without human explanation. ## DataHub MCP - [Official DataHub MCP repository on GitHub](https://github.com/qeasy-cloud/datahub-mcp-ops): The official open-source DataHub MCP implementation for code review, release tracking and collaboration. More Qeasy open-source projects will join the same GitHub ecosystem. - [DataHub Platform MCP](https://www.qeasy.cloud/en/skill/datahub): Operations MCP: 39 tools (meta.* / asset.* / strategy.* / script.* / audit.*) with tenant one-click authorization via sh-xxx keys (90 days by default), identity passthrough and full audit trails. Public cloud endpoint https://pro-service.qliang.cloud/mcp; private deployments use self-describing keys and multi-profile; install with npx skills add qeasy-cloud/datahub. Roadmap: strategy.delete / script.test / asset.connector.list / audit.stats and guarded human-in-the-loop are not implemented. - [Data Integration Knowledge](https://www.qeasy.cloud/en/mcp/datahub-knowledge): The data-integration domain of the unified knowledge MCP: K1 platform operations & maintenance / K2 platform APIs & metadata / K3 integration design patterns. knowledge_search accepts product=data-integration plus scenario (S1/S2/S3, default S2, hard-filtered before recall) and category (K1/K2/K3); citations are docs/knowledge/#
; SOL/CASE/FAIL/REF/TPL/BP layers stay restricted (the other nine layers are public). - [Reconciliation Knowledge](https://www.qeasy.cloud/en/mcp/reconciliation-knowledge): The reconciliation domain of the unified knowledge MCP: 12 knowledge domains, platform topics and customer cases, one atom per chunk with docs/knowledge/.md#KA-... citations; general methodology and standards are public. The reconciliation operations MCP is fully privately deployed; the website currently publishes only the knowledge base. ## Reconciliation Resources - [Pricing · Reconciliation Subscription and Perpetual](https://www.qeasy.cloud/en/pricing): E-commerce reconciliation pricing — subscription ¥12,000/year (buy 5 years get 1 free); perpetual ¥86,000 (one-time, fully on-prem); implementation pack ¥30,000 for 10 platform integrations, +¥6,000 each; 14-day free trial on subscription. - [Reconciliation Field Notes](https://www.qeasy.cloud/en/insights/reconciliation): Production-tested reconciliation playbooks: per-platform matching rules, variance-handling checklists, five-dimension accounting, cross-border specials and customer cases. ## Insights (Core Content) - [Insights — Qeasy Content Hub](https://www.qeasy.cloud/en/insights): Qeasy's core content hub: in-depth industry insights, integration tutorials, production best practices, customer cases and FAQ about data integration, ERP/e-commerce connectivity and AI agents. Bilingual and AI-search friendly with structured GEO data. - [Integration Solutions](https://www.qeasy.cloud/en/insights/solutions): End-to-end integration solution articles connecting ERP, e-commerce and WMS systems. - [Product Docs](https://www.qeasy.cloud/en/insights/product-docs): Platform product documentation and usage guides (TechArticle JSON-LD). - [Engineering](https://www.qeasy.cloud/en/insights/engineering): Engineering deep-dives: sync strategies, observability, idempotency and reliability patterns. - [Case Studies](https://www.qeasy.cloud/en/insights/case-study): Real-world customer integration case studies. - [FAQ](https://www.qeasy.cloud/en/insights/faq): Common integration questions with FAQPage JSON-LD for LLM citation. - [Glossary](https://www.qeasy.cloud/en/insights/glossary): Precise definitions of data-integration terms (iPaaS, ETL, CDC, idempotency) for humans and LLMs. ## Solutions - [Integration Solutions](https://www.qeasy.cloud/en/solution): Turnkey data integration for ERP, CRM, e-commerce, distributor and cross-border scenarios. - [Distributor Data Integration](https://www.qeasy.cloud/en/distributor-data-integration): Multi-distributor ERP integration for FMCG and food brands. - [Kingdee Cloud Integration](https://www.qeasy.cloud/en/solution/kingdee): Deep-dive topic page for Kingdee Cloud connectivity. - [Yonyou Integration](https://www.qeasy.cloud/en/solution/yonyou): Deep-dive topic page for the Yonyou product line. ## Resources - [Blog](https://www.qeasy.cloud/en/blog): Hands-on integration guides, API tutorials and case studies. - [Customer Cases](https://www.qeasy.cloud/en/case): Real-world system connectivity case studies. - [Contact Us](https://www.qeasy.cloud/en/contact): Solution consulting and technical support; apply for authorized MCP access here. - [Terms of Service](https://www.qeasy.cloud/en/terms): Account, service, data and billing usage rules. - [Privacy Policy](https://www.qeasy.cloud/en/privacy): What we collect, why, and how data is protected. - [HTML Sitemap](https://www.qeasy.cloud/en/sitemap): Human-readable site map of all major entry points. - [XML Sitemap](https://www.qeasy.cloud/sitemap.xml): Full machine-readable sitemap with all dynamic URLs. - [llms.txt](https://www.qeasy.cloud/en/llms.txt): This file — the English llms.txt site content map. Chinese version: /llms.txt. ## M9 Integrated Platform API Docs - [/api-v2/ Platform Directory](https://www.qeasy.cloud/api-v2): Directory of integrated platforms (168 ERP / CRM / e-commerce systems) with their official API documentation aggregated and indexed for search. ## M9 Knowledge Graph - [Knowledge Graph Explorer](https://www.qeasy.cloud/en/kg/explore): Visual exploration of entities (platforms, concepts, API endpoints, industry terms) and their relationships.