Wenbu 1.3.1 · Official organization and agent integrations

Wenbu now lives at wenbu-app/wenbu, with a bilingual organization home and direct links to the free tools, illustrated handbook and agent integrations.

This release updates the CLI package repository and MCP Registry source metadata to the new organization. The remote MCP endpoint remains https://wenbu.app/mcp; calculation behavior is unchanged. Existing GitHub links redirect, and earlier release archives are preserved.

问卜已迁入 wenbu-app 组织。组织主页集中介绍产品、知识手册与开源接入方式，官网中英文页面也已同步新源码地址。独立工具和知识内容免费；内置 DeepSeek Agent 受公开每日额度限制。计算与资料访问无需账户或 API key。

• Try Wenbu / 免费使用
• MCP quickstart, CLI, installable Agent Skill
• Official MCP Registry

Download the three archives below and verify their SHA-256 hashes with SHA256SUMS. The CLI requires Node.js 22.12 or later. No npm publication is claimed.

Validation: 282 tests, type checks, lint, site/content/art audits, and PR/main CI passed. These tools support cultural reflection; calculation accuracy does not establish predictive accuracy.

Wenbu v1.3.0 · Open agent integrations & calibrated usage

Wenbu · 问卜 is a free bilingual workspace for BaZi, the I Ching, tarot and Zi Wei, with a source-based research agent. These tools support cultural reflection; calculations are transparent, and interpretations are not established predictions.

Try Wenbu · English · Integration guide · First reproducible connection

Official MCP Registry record · app.wenbu/mcp v1.3.0. Domain-authenticated remote endpoint and public source metadata; directory publication does not imply endorsement or installation.

Connect your own agent

• MCP: Streamable HTTP at https://wenbu.app/mcp; six calculation and library tools; no Wenbu account or API key. Your host supplies its model. The MCP archive contains configuration, an official-registry manifest and a fixed-lines request example.
• CLI: Node.js 22.12+, no dependencies. Download the archive, verify SHA256SUMS, inspect it, then install the local tarball or run wenbu.mjs directly. This release does not claim npm registry publication.
• Skill: The installable skills/wenbu/SKILL.md explains tool selection, source verification and the context a person chooses to share. Installing it does not automatically configure MCP.

All artifacts are built from the public MIT source. Production credentials, conversations, feedback and private analytics are excluded. Use a named input file or stdin for personal details. The optional built-in DeepSeek Agent has published daily allowances and sends only the selected context.

Measurement improvements

MCP calls now receive one terminal receipt, including SDK validation failures. Overview and history share calendar ranges, dimensions and snapshots. The private insights dashboard separates browser identifiers, confirmed-use identifiers, content fetches and service calls; it exposes classification evidence, legacy coverage and collection diagnostics. QA markers are consistent across collectors and test traffic stays out of ordinary statistics.

Validation: 282 tests, type checks, lint, CI, 93 built pages / 86 indexable URLs, 42 complete guide editions, 78 card images, mobile/desktop UI, live asset matching and production MCP/report reconciliation. Grok CLI returned HTTP 402 and did not complete a review. Registry/indexing/traffic outcomes are reported separately; this release does not establish new user growth.

中文：这一版把 MCP、CLI、Skill 整理为可直接使用的开源接入包，同时校准浏览、抓取、调用与真实完成使用的统计口径。官网与教程中英文均可使用；所有测试记录默认排除，不把浏览器标识包装成自然人数。